SANDIA REPORT SAND85-0380· Unlimited Release. UC-70 Printed July 1985

Proceedings of the Workshop on the Source Term for Radionuclide Migration From High-Level Waste or Spent Nuclear Fuel Under Realistic Repository Conditions

T. O. Hunter, A. B. Muller, Editors .

Prepared by Sandia National Laboratories Albuquerque, New Mexico 87185 and Livermore, California 94550 for the United States Department of Energy under Contract DE·AC04·76DP00789

SF2900Q(S-S 1 )

When printing a copy of any digitized SAND Report, you are required to update the

markings to current standards.

Pursuant to article I of the Convention signed in Paris on 14th December, 1960, and which came into force on 30th September, 1961, the Organisation for Economic Co-operation and Development (OECD) shall promote policies designed:

to achieve the highest sustainable economic growth and employment and a rising standard of living in Member countries, while maintaining financial stability, and thus to contribute to the development of the world economy; to contribute to sound economic expansion in Member as well as non-member countries in the process of economic development; and to contribute to the expansion of world trade on a multilateral, non-discriminatory basis in accordance with international obligations.

The Signatories of the Convention on the OECD are Austria, Belgium, Canada, Denmark, France, the Federal Republic of Germany, Greece, Iceland, Ireland, Italy, Luxembourg, the Netherlands, Norway, Portugal, Spain, Sweden, Switzerland, Turkey, the United Kingdom and the United States. The following countries acceded subsequently to this Convention (the dates are those on which the instruments of accession were deposited): Japan (28th April, 1964), Finland (28th January, 1969), Australia (7th June, 1971) and New Zealand (29th May, 1973).

The Socialist Federal Republic of Yugoslavia takes part in certain work of the OECD (agreement of 28th October, 1961) .

The GECD Nuclear Energy Agency (NEA) was established on 20th April, 1972, replacing GECD's European Nuclear Energy Agency (ENEA) on the adhesion of Japan as a full Member.

NEA now groups all the European Member countries of GECD and Australia, Canada, Japan, and the United States. The Commission of the European Communities takes part in the work of the Agency.

The primary objectives of NEA are to promote co-operation between its Member governments on the safety and regulatory aspects of nuclear development, and on assessing the future role of nuclear energy as a contributor to economic progress.

This is achieved by:

encouraging harmonisation of governments' regulatory policies and practices in the nuclear field, with particular reference to the safety of nuclear installations, protection of man against ionising radiation and preservation of the environment, radioactive waste management, and nuclear third party liability and insurance; keeping under review the technical and economic characteristics of nuclear power growth and of the nuclear fuel cycle, and assessing demand and supply for the different phases of the nuclear fuel cycle and the potential future contribution of nuclear power to overall energy demand; 'i

developing exchanges of scientific and technical information on nuclear energy, particularly through participation in common services; setting up international research and development programmes and undertakings jointly organised and operated by GECD countries.

In these and related tasks, NEA works in close collaboration with the International Atomic Energy Agency in Vienna, with which it has concluded a Co-operation Agreement, as well as with other international organisations in the nuclear field.

Issued by Sandia National Laboratories, operated for the United States Department of Energy by Sandia Corporation. NOTICE: This report was prepared as an account of work sponsored by an agency of the United States Government. Neither the United States Govern· ment nor any agency thereof, nor any of their emploiees, nor any of their contractors, subcontractors, or their employees, makes any warranty, ex­press or implied, or assumes any legal liability or responsibility for the accuracy, completeness, or usefulness of any information, apparatus, prod­uct, or process disclosed, or represents that its use would not infringe privately owned rights. Reference herein to any specific commercial product, process, or service by trade name, trademark, manufacturer, or otherwise, does not necessarily constitute or imply its endorsement, recommendation, or favoring by the United States Government, any agency thereof or any of their contractors or subcontractors. The views and opinions expressed here­in do not necessarily state or reflect those of the United States Government, any agency thereof or any of their contractors or subcontractors.

Printed in the United States of America Available from National Technical Information Service U.S. Department of Commerce 5285 Port Royal Road Springfield, VA 22161

NTIS price codes Printed copy: A 10 Microfiche copy: AD1

SAND85-0380 Unlimited Release Printed July 1985

Distribution category UC-70

PROCEEDINGS OF THE WORKSHOP ON THE SOURCE TERM FORRADIONUCLIDE MIGRATION FROM HIGH-LEVEL WASTE OR

SPENT NUCLEAR FUEL UNDER REALISTIC REPOSITORY CONDITIONS

Albuquerque. NM. USA November 13-15. 1984

Edited by

T. O. Hunter. Technical Chairman Sandia National Laboratories

A. B. Muller. Technical Secretary OECD Nuclear Energy Agency

Nuclear Energy Agency Organisation for Economic Co-operation and Development

Hosted by US Department of Energy

3

ACKNOWLEDGMENT

The editors wish to thank all the participants for the stimulating and revealing information provided in the papers and the discussion during the workshop. We are especially indebted to Dr. Ivars Neretnieks, Dr. M. J. Apted, and Dr. Ian G. McKinley for their perceptive summaries of the three technical sessions. (These summaries are provided with the papers in the proceedings.)

We wish to thank Ms. Renae Coleman of the U.S. Department of Energy for her support and dedication in the preparation for the workshop.

Finally, we must acknowledge the splendid and efficient service of Ms. Susan Grodin, who provided the communications and administrative support that did so much to ensure the success of the workshop.

4

CONTENTS

INTRODUCTION

SESSION I

SUMMARY OF SESSION I: ROLE AND USE OF SOURCE TERM

SESSION I PAPERS

1. Everett A. Wick How Reliable Does the waste Package Containment Have to Be?

2. L. D. Rickertsen and A. Van Luik Integration of Repository and Waste Package Performance in

the Evaluation of Source Terms

3. N. Stelte and R. Storck The Importance of the Source Term for the Release of Radio­

nuclides from a Repository in a Salt Dome

4. Ivars Neretnieks Source Term Modelling in the Swedish KBS-3 Study

5. J. W. Braithwaite and M. S. Tierney Source-Term Considerations for a Potential Radioactive-waste

Repository Located in Unsaturated Tuff

6. Gilbert E. Raines and John F. Kircher Data Requirements Based on Performance Assessment Analyses of

7

9

11

17

29

41

69

Conceptual Waste Packages in Salt Repositories (Abstract only) 81

SUKKARY OF SESSION II: CONTROLLING MECHANISMS OF RELEASE

SESSION II PAPERS

1. E. Peltonen. M. Snellman. and K. Ollila Release Mechanisms and Their Role in the Spent Fuel Repository

Performance

2. T. F. Rees. J. M. Cleveland. and K. L. Nash Leaching of Plutonium from a Borosilicate Glass by Selected

Ground Waters

5

83

87

103

CONTENTS (Continued)

3. I. G. McKinley The Quantification of Source-Te~ Profiles from Near-Field

Geochemical Models 115

4. M. J. Apted, D. H. Alexander, A. M. Liebetrau, A. E. Van Luik, R. E. Williford, and P. G. Doctor

A Conceptual Model for Repository Source-Te~ Evaluation 125

SUMMARY OF SESSION III: EXPERIMENTS AND SPECIAL EFFECTS 137

SESSION III PAPERS

1. M. N. Gray, R. B. Heimarin, J. C. Tait, and A.G. Wikj ord System Tests for Studying the Release of Radioactivity from High-

Level Waste Fo~s under Geological Disposal Conditions 139

2. F. Lanza,E. Parnisari, and C. Ronsecco Study of the Release of Various Elements from HLW Glasses in Con-

tact with Porous Media 151

3. A. Avogadro, G. Bidoglio, and A. Saltelli Laboratory Simulation of Radionuclide Geological Migration 163

4. V. M. Oversby and R. D. McCright Laboratory ExPeriments Designed To Provide Limits on the

Radionuclide Source Te~ for the NNWSI Project 175

5. G. L. McVay, L. R. Pederson, W. J. Gray, and D. Rai The Potential Effects of Radiation on the Source Term in a

Salt Repository (Abstract only) 187

6. D. W. Shoesmith and M. G. Bailey The Chemistry of Nuclear Fuel (U02) Dissolution under Waste

Disposal Vault Conditions 189

WORKSHOP SUMMARY AND CONCLUSIONS 201

PARTICIPANTS LIST 203

6

INTRODUCTION

Inherent in the creation of energy from nuclear power is the generation of some nuclear wastes. The management and disposal of these wastes is a burden accepted along with the benefits derived from nuclear energy. Deep geologic disposal is preferred in most NEA countries as the disposal option for high-level nuclear waste or for spent nuclear fuel. It is current technical opinion within the NEA Member countries (NEA, 1985) that the safe disposal of such wastes deep in stable continental geologic formations appears to be feasible on the basis of existing technology.

The assessment of the performance of these disposal systems and the demonstration of their long-term safety relies heavily on modeling the behavior of the engineered repository and the surrounding host geological medium as an integrated waste isolation system (NEA, 1983). This involves a detailed description of the system of natural and engineered barriers to radionuclide migration into the biosphere and an evaluation of the perfor­mance of this system under a number of plausible scenarios. Essential to the modeling is the ability to predict the quantities, rates, and pathways in which radionuclides released from a repository may migrate in the geo­sphere. This in turn depends on a number of natural and engineering param­eters. Among the parameters to which the modeling results are most sensi­tive is the "source term," which describes the rate at which the various radionuclides are mobilized from the waste.

At this point, there is no universally accepted view as to where in the system the "source term" refers. Although regulatory or conventional definitions exist, it is generally felt that from a technical modeling per­spective, the source term refers to the specification of the flux of radio­nuclides across the interface between the engineered repository model and the geosphere transport model. One of the purposes of this workshop was to examine the role and appropriate definition of the source term for use in modeling.

until recently, performance analyses usually involved making major assumptions about this term. These assumptions were based on the calculated radionuclide inventory in the repository at a given time, discharging at an estimated release rate. The rate was determined based on assumptions about canister lifetime and waste form stability. since performance analysis results have been shown to be extremely dependent on this term, it has become necessary to assess the realism of the assumptions and to develop more detailed phenomenological models of the source term.

Such model development requires evolving a more fundamental understand­ing of the physical and chemical processes that influence the source term. Experiments under realistic in situ conditions are essential to this under­standing. A number of the papers in these proceedings describe such experi­ments. Identifying and quantifying the effect of parameters that most

7

influence source term are also essential. This task appears throughout the papers in this volume. Finally, the role of source term modeling .in the broader perspective of repository system performance assessment is discussed in several of the papers. Together they are an attempt to summarize the current state of understanding of the source term and of its modeling under realistic repository conditions.

8

SUMMARY OF SESSION I: ROLE AND USE OF SOURCE TERM

An issue underlying the assessment of the role and the use of radio­nuclide source terms in performance assessment modeling is the question of where the "release" takes place, i.e., at what point is the "source term" to be defined. Is it at the outside of the waste package, further out in the near field, or at the near-field/far-field boundary. The point at which the source term is defined may, in some cases, affect the type of model or model coupling to be used. The question is of practical, as well as regulatory, importance. Although the question was not resolved, it seems to have been generally felt that in practical, modeling terms, the source term should be defined at the interface where the waste package model is coupled to the next model (be it a near-field model for HLW or a far-field transport model for LLW or ILW) in the overall performance assessment scheme used.

A number of important specific points on the definition and role of the source term were made by speakers in this session. It was shown (Rickertson) that the physical relationship to and the influence on each other of the waste packages in a repository may considerably influence and reduce the release from an individual package. The geometry of the repository and the direction of water flow through it can considerably influence the release rate in cases where the release from individual canisters is of short dura­tion. The spread in time of degradation of the waste packages can also give a considerable reduction of the release rate from the whole repository.

The presentation on releases from a salt repository (Stelte) showed that because of the present lack of rate data on corrosion and rate data on concrete interaction with the water, stacked conservative assumptions had to be used to assess release rates. The small amount of available water had to be assumed to fully participate in both reactions, whereas in reality, one mechanism would dominate and there would be less water available to the other reaction. The stacking of conservative assumptions was also observed (by Neretnieks) to give strongly exaggerated release rates in a granite repository. It was generally recognised that more realistic modeling will give a considerable reduction in projected release rates when better data become available. An example where many different mechanisms such as con­gruent release rate, solubility limits, decay in backfill, and changes of redox conditions due to alpha radiolysis, were integrated in a release model was demonstrated (Neretnieks). The redox front was projected to extend out into the rock outside the backfill. It was contaminated within a few tens of meters from the canisters. In rock with less ferrous iron minerals, the oxidized region might extend much farther and impact the far-·field condi­tions.

The excavation of the repository was shown (Braithwaite) to have a beneficial effect on the flow rate of water to the waste package because of the high capillary suction pressures of the tuff in this case. The unsatu­rated tuff will not give off its water to an empty hole. In this situation,

9

a backfill with higher capillary suction will draw in more water than if the backfill were designed for a lower capillary suction than the rock. It was also projected that in an unsaturated medium, the low availability of water may only wet a very small fraction of the waste surface, thus decreasing the contact area and leach rate. As several others did, this presentation used sensitivity analysis as a tool to identify the sensitive mechanisms and parameters.

The need for site-specific data in any detailed source-term modeling was emphasized by all speakers. The question of gas production and escape, both from radiolysis and from generation during canister corrosion, was generally felt to be an area warranting additional work.

10

HOW RELIABLE DOES THE WASTE PACKAGE CONTAINMENT HAVE TO BE?

Everett A. ~Iick U.S. Nuclear Regulatory Commission

Washington, DC 20555

ABSTRACT

The final rule (10 CFR Part 60) for Disposal of High-Level Radioactive Wastes in Geologic RepositoY';es specifies that the engineered barrier system shall be designed so that, assuming anticipated processes and events, containment of high-level radioactive wastes (HLW) will be substantially complete during the period when radiation and thermal conditions in the engineered barrier system are dominated by fission product decay. This requirement leads to the Nuclear Regulatory Commission (NRC) being asked the following questions: What ;s meant by "substantially corrplete"? How reliable does waste package containment have to be? How many waste oackages can fail? Although the NRC has not defined quantitatively the term "substantially complete", a numerical concept for acceptable release during the containment period is discussed. The number of containment failures that could be tolerated under the rule would depend upon the acceptable release, the time at wrich failure occurs and the rate of release from a failed package.

11

1.0 INTRODUCTION

Under the Energy Reorganization Act of 1974 the Nuclear Regulatory Commission (NRC) has responsibility to lic~nse the disposal of high-level nuclear waste (HLW) by the Department of Energy (DOE) [1]. In addition, the Nuclear Waste Policy Act of 1982 states that it is federal responsibi.lit~ to provide for HLW and spent fuel requiring permanent disposal [2j. Thus, DOE has responsibility for disposal of HLW and spent fuel and NRC has responsibility to license the DOE repository or repositories.

The regulation for Disposal of High-Level Radioactive Wastes in Geologic Repositories, 10 CFR Part 60, specifies [3J that the engineered barrier system shall be designed so that, assuming anticipated processes and events, containment of HLW (within the waste packages) will be substantially complete during the period when radiation and thermal conditions in the engineered barrier system are dominated by fission product decay (300 to 1000 years).

The regulation also specifies [3] that the release rate of any radionuclide from the engineered barrier system following the containment p~riod shall not exceed one part in 100,000 per year of the inventory of that radionuclide calculated to be present at 1000 years following permanent closure, or such other fraction of the inventory as may be approved or specified by the Commission.

The requirement for substantially complete containment within the waste packages leads to the· NRC being asked the following questions:

1. ~lhat is "substantially complete"?

2. How reliable does the waste package containment have to be?

2.0 DISCUSSION

2.1 WHAT IS "SUBSTANTIALLY COMPLETE?"

NRC has not defined quantitatively the term "substantially complete containment". DOE, however, could consider the following or other appropriate approaches:

Limit the release of radionuclides from th~ waste packages during the containment period to the same number of curies that ;s permitted annually from the engineered barrier system following the containment period.

12

2.2 HOW RELIABLE DOES WASTE PACKAGE CONTAINMENT HAVE TO BE?

If the term IIsubstantially complete ll were interpreted quantitatively as described above, the level of reliability required for waste package containment could be defined as follows:

Waste package containment should be sufficiently reliable so that radionuc!~de releases during the containment period do not exceed 1 x 10 per year of the radionuclide inventory calculated to be present 1000 years after permanent closure.

The reliability of the waste package that would be required during the containment period would be such that the calculated cumulative number of failures during the containment period and the release calculated to result from them would not exceed the quantity stated above.

A basis for calculating the number of waste package failures that could be tolerated during the containment period is presented in Section 2.3.

2.3 HOW MANY WASTE PACKAGES CAN FAIL?

The number of waste packages that can fail during the containment period without exceeding the release criterion depends upon:

1.

') '- .

The estimated fractional rate of release from a failed package.

When failure occurs.

Although the number of curies that may be ~5leased annually during the containment period ;s constant (lx10 ' x 1000 year inventoryL the acceptable fractional annual release will increase with time to failure. It may be calculated as follows:

Acceptable Fractional Annual Release =

1 x 10-5 x 1000 r. inventor current inventory

For example~ the total activity in curies of inventory in one BWR spent fuel rod varies with time after discharge as shown below [4J:

10 years 960

AOO years 11

1000 years 4.6

This example is also true on a relative basis of the total inventory of spent fuel waste packages in the repository at 10, 300,

13

and 1000 years. Therefore, the permissible annual release at 10 years after emplacement (in this example, the time between discharge and permanent closure was taken as zero, although we know this will not be the case) may be calculated as a fraction of the total radionuclide inventory at that time, e.g.,

4.6 curies x 1xl0-5 x inventory 10 years after 960 curies permanent closure

= 4.8 x 10-8 of current (10 yr.) inventory.

Once the acceptable annual fractional release is known and the annual release from a failed waste package is estimated, an acceptable number of waste package failures may be calculated.

Assumptions:

1. The repository will contain 70,000 waste packages.

2. A failed waste package will release 1 x 10-5 per year of the inventory in the package.

Thus, the fractional annual rate of radionuclide release from one failed waste package would be:

number of curies 1 x 10-5/yr. x package = 1.43 x 10-10/yr . -n-um-;b;-e-r-o"';=f~cu-r-'~· e-s--x--.7o;";O .. ,-?;0"'O""O-pa ..... c'k-a-g-e s

package

Therefore, the number of waste package failures that could be tolerated 10 years after permanent closure in a repository containing 70,000 wastes packages is:

acceptable fractional release (4.8 x 10-8/yr) = 335 packages

fractional release per package (1.43 x 10-10/yr)

Therefore, 335 package failures could be tolerated 10 years after permanent closure ~f annual fractional release of a failed package does not exceed 1 x 10- of the inventory in the package. The release rate fro~ a failed package at 10 years, however, may be much higher than 1 x 10- per year because the temperature will be relatively high. For example, if the leak rate were a hundred times hig~er only three failed packages could be tolerated; if it were a thousand times higher, none could be tolerated.

3.0 CONCLUSIONS

1. NRC has not defined c;uantitatively the term IIsubstantially complete containment. 1I One of the approaches that is being considered, however, is to allow a radionuclide release rate

14

during the containment period that does not exceed the absolute qu~§t;ties permitted in the post-containment period, i.e., 1 x 10 per year of the radionuclide inventory of the repository 1000 years after permanent closure. Since the radionuclide inventory is larger during the containment period, the fractional release at the time of containment failure must be correspondingly smaller.

2. If such an approach were allowed, the reliability of the waste package containment should be such that the calculated cumulative number of failures during the containment period and the releases calculated to result from them would not exceed the quantity stated above.

3. The cumulative number of waste package failures that could be tolerated during the containment period would depend upon when failure occurred and the rate of radionuclide release from a failed package.

REFERENCES

1. Public Law 93-438, Energy Reorganization Act of 1974, Sec. 202, 42 U.S.C. 5842.

2. Public Law 97-425, The Nuclear Waste Policy Act of 1982, 96 stat. 2207, Sec. 111(a).

3. 10 CFR 60, "Disposal of High-Level Radioactive Wastes in Geologic Repositories," Subpart E, Technical Criteria~ paragraph 60.113.

4. NUREG/CR-2482, Vol. 7, "Review of DOE Waste Package Program, Subtask 1.1 - National Wa~te Package Program, Draft Biannual Report ll

, Evelyn Gause, Peter Soo, September 1984, pp. 31 (adapted from ORIGEN-2 calculated values presented in ORNL/TM-6008, 1977).

15-16

INTEGRATION OF REPOSITORY AND WASTE PACKAGE PERFORMANCE IN THE EVALUATION OF SOURCE TERMS

L. D. Rickertsen and A. Van Luik Roy F. Weston lnc./Rogers and Associates Engineering Corp.

2301 Research Blvd. Rockville, Maryland 20850 (USA)

ABSTRACT

A number of features of the near-field regime, in addition to the waste form and waste package release processes, may be important to determination of the source term. These aspects include waste package degradation rates and the spatial distribution of waste package sources through the repository. This paper considers the relative importance of these features in a simple but useful model. The sensitivity of the source term to these features is explored for representative geologic systems currently being evaluated in the U.S. program.

17

INTRODUCTION

Performance assessments of geologic repository systems that involve modelling of transport of radionuclides from the repository through the hydrogeologic system to the accessible environment generally do not rely upon detailed representations of the source term. For example, the repository release is approximated by a single block or point source with a release rate estimated based on waste package or waste form analyses. However, because there will be many waste packages in the repository and because these waste packages will be distributed throughout a large section of the host rock, dispersive effects can occur within the repository that may significantly affect the repository releases to the far field relative to the waste package release rates. Not all such effects can be evaluated without detailed site and repository data. However some of the effects can be estimated from relatively rudimentary considerations. It is the purpose of this paper to investigate the feasibility and advisability of incorporating some of these simple considerations into the current repository system performance assessment efforts.

GEOMETRIC EFFECT

The first feature of the repository that is considered is the distribution of waste packages throughout the repository. The contribution of different waste packages to the release from the repository will depend upon the relative position of the waste packages within the repository. A simple example of this geometric effect is shown in Figure 1. The large peak represents the fractional rate of release of the repository inventory at the repository boundary neglecting the spatial separation Of the waste packages. This peak is simply the direct sum of the release rates from all packages independent of their location in the repository. In reality, however, the releases from the packages located near to this boundary would reach the boundary sooner than those farther away. Therefore, at a given time the contribution from each waste package to the source term depends upon the distance from the boundary and the time needed for radionuclides to reach that

AU. CANtST£1IS AT A SllIOlE lOCAT1ON

Figure 1. Effect of Spatial Distribution of Waste Packages.

18

boundary. In the simple picture in Fig·ure 2, the large peak is decomposed into a series of smaller peaks, effectively reducing the peak release rate and spreading the release out in time. For a repository with as many as 30,000 waste packages, such modifications could be significant.

The impact of the geometry is a function of the boundary chosen for the evaluation of the repository release rate. The choice of this boundary depends upon the radionuclide pathways through the repository in the scenario being considered. Since the rate of release at this boundary is dependent upon the flux of radi0nuclides across the boundary, the geometric effect is also a function of the radionuclide velocity at this boundary.

The effect also depends upon any dispersion in the transparent of the radionuclides through the repository before reaching the boundary. Figure 2 shows how the release rate at the boundary can be affected •. In this case the source term is effectively averaged over the individual waste package release rates. One possible source of such an effect is hydrodynamic dispersion of the radionuclide concentration within the repository. In this case, the fractional release. rate due to the distribution of waste throughout the repository can be approximated by

q(t) = V L

~ [ erf (Xo + L - Vt) -y4aVt

erf (Xo - Vt>1 y4aVt

where erf is the error ·function, V is the radionuclide velocity, L is the length of the repository, Xo is the distance of the repository from th~ boundary, and a is the dispersivity. This expression is illustrated in Figure 3. Although this representation for the geometric effect has been derived from simplistic hydrodynamic dispersion considerations, it applies to any of the dispersive effects that may occur in the repository. In general, the effect of the geometry on the source term will be to·provide a release duration L/V, corresponding to the (retarded) transit time of the radionuclide the repository, and an average fractional release rate of V/L.

The integration of the ·waste package and repository performance requires some care. The release at the boundary at a specified time may have contributions from more than one waste package. For example, radionuclides wnich take a long time to travel to the boundary may arrive at the same time as those released later from waste packages nearer to the boundary. The total release rate is evaluated by integrating over all such contributions. Formally, this integration is the convolution over the waste package fractional release rate, qp' and the geometric term, qR:

The integration is illustrated for a simple representation of the waste package release rate in Figure 4. The fractional release rate from the combined system has a behavior which differs significantly from the single-packaged behavior.

19

i ! i ... c 5

i

I I

IINOLI PACKAGE DlII'EIIIIOM

Figure 2. Effect of Dispersion on Geometric Release Rate.

"'V LIV

tIME

Figure 3. Analytic Solution for Geometric Effect.

20

UNIPOIIM LIACH IIATI

I'T,I------....,

T,

i O!OMETIIY

i I I'T,I-___ ----------...

T,

COMBINED "ATE

"T~

T,

Figure 4. Integration of Leach Rate and Geometry.

§

i I WAITE 'ACKAO! FAlLUllEI DIITIUMITED

-Figure 5. Effect of Waste Package Failure Rate.

21

The peak value of the integrated source term corresponds to the smaller of qp and qR' This result makes sense physically and, in fact, it can be shown from the properties of the convolution integral that the fractional rate of release from the combined system is always less than or equal to the smaller of the component fractional release rates. Therefore, it is possible that the source term that accounts for the repository geometry can be much smaller than that evaluated from the waste package analyses alone.

WASTE PACKAGE FAILURE RATE

A second feature of a repository that can influence the source term is the fact that waste package failure is a stochastic process in which the failures are distributed over time. Since no release is possible until the waste package fails, a distribution of the failure rates can affect the overall release rate. Figure 5 compares the fractional release assuming all packages fail at the same time to the rate in which the failures occur distributed over time. This figure does not take into account the geometric distribution of waste packages. For the case where the distribution of failure is taken into account, not only is the release rate spread over a longer time but, because the releases at a given time come from a subset of the total set of waste packages, the peak release rate is reduced. Therefore, the integration of the waste package failure rate into the waste package release rate may affect the determination of the source term.

Figure 6 shows the result when the waste package release rate is integrated over a probability distribution for the failure rate. A single-parameter, Poisson-type failure rate is assumed in this case and the result shows that a significant effect on the system release rate is possible.

INTEGRATION OF REPOSITORY AND WASTE PACKAGE PERFORMANCE

Figures 7 and 8 show the effect of taking ihto account both the geometric effect and the waste package failure rate in evaluating the source term. Figure 7 shows the impact of the geometric effect on a system in which the leach rate is one part in lOs per year and for which the Poisson failure frequency is 10- 4 per year. For a repository transit time of 10,000 years (fractional release rate of 10- 4 per year), the only detectable effect is in the first 20,000 years and in the tail of the distribution. For a repository transit time of 10,000 years, the effect is more pronounced. Finally, for a repository transit line of 1,000,000 years, the source term is dramatically reduced by the geometric effect.

The sensitivity of the calculated source term to the waste package failure rate is illustrated in Figure 8. In this case the repository transit time is set at 1,000 years and the waste package release rate is set to 10- 4

yr- 1 For a failure frequency of 10- 3 yr- 1 there is some effect on the source term. This effect becomes much more important as the frequency is reduced below 10- 4 yr -1.

In these examples, the peak value of the source term is controlled by the most constraining of the three effects. For 'early times, the peak value can be much less than the peak value from any individual component due to synergistic effects.

22

Figure 6.

UNIFORM LEACH RATE Q • 1IT,

T,

POISSON FAILURE Q ;; l', e.p (-n.t)

,f"

lIT~

TIME

Integration of Leach Rate and Waste Package Failure Rate.

TIME ('_ VAS)

Figure 7. Dependence of Source Term on Geometric Effect.

23

APPLICATIONS TO SPECIFIC SITES

Although it is still too early to perform definitive analyses for any specific site, existing information can be used to investigate the relative importance of the effects being discuSS9Q. Figure 9 shows the results calculated for a repository constructed in a ba~alt flow. Such a formation is presently being in~estigated ~t the Hanford Site near Rich~and, Washington. Preliminary evaluations indicate that the expected time to failure for a waste package under these conditions may be as high as 6500 Years with a variance of 2000 years based upon existing corrosion data. A Gaussian distribution that gives these val~es has been a~sumeq for the failure rate. The waste package release rate is based upon leach rate data for spent fuel and a leach rate of 2 x 10- 5 yr- I is used here. Radionuclide solubility limits to the leaching are not considered here. The scenario envisioned is one in which the ground water moves up vertically thro~gh the repository, dissolving radionuclides leached from the waste form. The ground wa~er rises to the flow top and the radionuclides are then carrieq horizontally along the flow top to the accessible environment. For a vertical velocity of 3 x 10~4 m/yr in the host rock and a repository thickness of one meter (based upon one meter diameter canisters emplaced horizontally), the geometric factor alone is 3.3 x 10- 5 yr- 1

• No radionuc1ide retardation is taken into account in this simple example.

Figure 9 shows that the waste package failure rate has some effect and delays the release by several thousand years. The integratiqn of the waste package release rate and the geometric effect produces a SOUrce term that peaks at about 50,000 years. While the peak releas~ rate is not substantially different from the leach rate, the source term in the f~rst 10,000 years is significantly reduceq. Therefore, estimates of releases to the accessible environment in this period co~ld depend upon reposito~y features.

Figure 10 shows the source term for a simple simulation of the ~ucca Mountain site in Nevada. A horizon in the unsaturated tuffs is being considere9 as ~ candidate for a repository at this site. D~e to the unsaturated conditions and low Qround-wqteF flux rates for this site, extremely low release rates are expected. For exam~le, the waste form dissolution rate is probably ~ess than 5 x lO-9 yr - due to the low flow rates. A geometric term of 10- 3 yr- 1 has been assumed based on a flux rate of about 3 x 10~4 m3/yr/m Z and a waste package length of about 3 meters. An exponential distribution function with a mean time to failure of 10,000 years and a variance of 7000 years has Qeen assumed. The results in Figure 10 show that the geometric term is of little consequence for this case since the waste package release rate is at least 5 orders of magnitude smaller than this effect. However, the release rate in the first 10,000 years is significantly reduced by the low failure rate.

The possible impact on source term for scenarios involving a disruption to the expected repository performance is shown in Figure 11. In this case a spent fuel repository in a bedded salt site is investigated. It is assumed that an exploratory borehole connects aquifers that overlie and underlie the host salt bed. Water flowing through the borehole can then dissolve the host salt. AS the salt dissolves, waste packages can be exposed to the water and

24

Figure 8.

I'~ • 10.1 V,,·,

nlo1. 11000 YAS)

~. to· I V".' ~, ., to.4 V".I

Dependence of Source Term on Failure Rate.

! i !

I ! GAUSSIAN W",STl! PACKAGI ; FA't.U"E "ATE

t.-; ; .• $00 YA

:: n - 2.000 YA

-·I------·-···~ -."" ~ ....... '"~. : i : { LEACH AATE q - bl0·· YA·'

···1t··· .. ················t··· .. · .. i i

1\ J \

I ' !

40 .0 100

Figure 9. Integrated Source Term for Basalt Site.

25

r\

'1\ :

\

\/ EXPONENTIAL WASTE PACKAGE FAILURE RATE

~ 0 10.000 YEARS (I " 7.000 YEARS

. .

\ :

\ \ \ \ \ \~

\\ '" .............

20

GEOMETRY q 0 10.1 YR.1

LEACH RATE" S.10·' YA·-

SOURCE TERM

30 co 50

TIME 11.000 YEARS)

Figure 10. Integrated Source Term for Yucca Mountain.

waste package failure can permit the leaching and transport of radionuclides from the repository. The geometric term therefore depends upon the rate at which the dissolution front exposes the packages. This exposure rate has been evaluated in detailed assessments of this scenario [1] and the results are shown in the insert in Figure 11. The calculated waste package exposure rate continues for only 8 years until saturation of the brine limits further salt dissolution. A leach rate of 2.2 x 10- 4 yr- 1 is assumed for the spent fuel and a Gaussian failure rate with a mean time to failure of 10,000 years and a variance of 2000 years have been assumed for the waste packages. The impacts of both the failure rate and the geometric term on the source term are dramatic. The peak release rate is reduced by more than an order of magnitude from the leach rate. The total calculated release from the reposi to·ry system in the first 10,000 years would therefore be quite limited although specific system assessments using this source term would be needed to verify this conclusion.

26

WASTE PACKAGE: ULIOSUAE: "An: DUE TO SALT DISSOLUTION 4'FTlR IOREHDLl INTRU5ION Of HDOIO SALT "ELlOSlTOA.,

LEACH RATE =: 2.2. '0·· YA·'

······· .. ···········v

, 4

! i I I ! :

..... ' ... ! \ GAUSSIAN WASTE PACKAGE i \ FAILURE RATE • l / -! Y , =: 10,000 VIARS

I '\ a = 2,000 VEARS i \ I \ i \ i \

/ \. i i I i I \

/ \. SOURCE TERM

l .A··

12 " 20

TIME ('000 YU"I)

Figure 11. Integrated Source Term for Borehole Intrusion at Bedded Salt Site.

CONCLUSION

These applications demonstrate that there may be cases where the integration of repository and waste package performance may have profound impacts upon source terms. In some cases the geometric effect may be very important as in the case when the transit time through the host rock is very long. It also appears that waste package failure rates should be taken into account. The key parameters of the failure rate include the delay before release or the mean time to release and the peak frequency of waste package failures.

27

It should be noted that there are many other effects that could affect the source term. For example, very-near-field transport mechanisms and host rock/water interactions are likely to provide significant impacts on system performance. The simple model described here does not take ~nto account these effects. Nor does it account for the dependence of the waste package release upon the location within the repository or the thermal and chemical conditions that may vary across the repository. ~hese effects should be considered in future site-specific evaluations.

REFERENCE

[1] Preliminary Analyses of Scenarios for Potential Human Interference for Repositories in Three Salt Formations, ONWI/E512-02900/TR-33, INTERA Environmental Consultants Inc., 11999 Katy F~eeway~ Suite 610, Houston, Texas 77079.

28

THE IMPORTANCE OF THE SOURCE TERM FOR THE RELEASE OF RADIO­NUCLIDES FROM A REPOSITORY IN A SALT DOME

N. Stelte and R. storck Gesellschaft fur strahlen- und Urnweltforschung mbH Munchen

Institut fur Tieflagerung D-3300 Braunschweig

Federal Republic of Germany

ABSTRACT

The release of radionuclides from a backfilled and brine satu­rated repository in a salt dome is analyzed as a function of canister lifetimes and solubility limits. The relationship be­tween the waste package and the multi-barrier system of the repository is discussed.

29

Introduction

In the West German "Projekt Sicherheitsstudien Entsorgung" (PSE) a safety analysis for the isolation of radioactive wastes in a salt dome repository has been carried out [1]. Some of the results concerning the influence of the source term are discussed in the following.

During an operational period of the repository of 50 years mainly wastes from nuclear power plants, corresponding to a total electric energy of 2500 GWa, will be deposited. Addi­tional wastes come from research, medicine and other sources. Although two disposal options - reprocessed and spent fuel - , both of which require different repository designs, have been studied within the project, for the sake of clarity, we shall concentrate on the case of the reprocessed fuel repository on­ly in the following.

The Repository Design

The site selected fot the repository is the salt dome at Gorleben. The dome is covered with sediments of 300 m thick­ness. The drift system of the repository lies at about 800 m below surface. A brief sketch is given in Fig. 1. Here the re­pository is reduced to representatives for the different types of waste locations. On the left side of the repository, dis­posal chambers contain low- and medium level wastes without heat production, while on the right side heat producing medium and high level wastes are stored in boreholes of 300 m depth measured from the drift level.

The Scenario

A scenario which can lead to the transport of radionu­clides into the biosphere has been selected to demonstrate the methodology developed so far. As the internal geological structure of the salt dome is yet unknown, it is assumed that a main anhydrite layer exists which forms a pathway for groundwater from the aquifer system of the overlying strata down to the repository level. It is conceivable that the shafts or nearby drifts at the center of the repository cross the anhydrite layer. After closure and sealing of the reposi­tory water may intrude into the repository from the main anhy­drite. In contrast to possibly eXisting brine pockets in rock salt which might shed a limited amount of brine into the re­pository, water intrusion from the main anhydrite is limited by the void space of the repository only. It is assumed that unsaturated intruding water becomes saturated brine upon con­tact with the rock salt formation.

The inflow of brine driven by hydrostatic pressure of about 10 MPa is slowed down by several engineered barriers,

30

W t-'

tumble­down chamber (TD-Ch)

Figure 1

top loading chamber (TL-Ch)

remote stacking chamber (RS-Ch)

/

brine from main anhydrite

MLW bore-hole (MLW-B)

HLW bore-hole (HLW-B)

Reprocessed fuel repository 120 years after the onset of brine intrusion.

i.e. the backfill material (crushed salt) in drifts, dams (va­rious materials) separating different parts of the repository, and sealings (salt concrete or condensed salt backfill) be­tween drifts and chambers or boreholes. Dams and sealings are represented by black areas in Fig. 1. The flow resistance of the anhydrite is neglected because of its unknown properties. The void space of the repository which can be filled up by the intruding brine decreases with time. This is due to the creep of rock salt which leads to convergence of drifts, chambers and boreholes in the repository. This process results also in the compaction of the backfill materials in the drifts thus increasing the flow resistances and stretching the inflow of brine in time. In this way convergence works against brine in~ trusion. However, when a drift is flooded convergence loses the competition in this part. The hydraulic pressure rises in that area and the convergence rate drops significantly.

As a flashlight Fig. 1 also shows the distribution of brine in the repository 120 years after the beginning of brine intrusion (and closure of the repository). Although a part of the drifts is already flooded, no contact between wastes and brine has taken place as yet. On the contrary, an important change of the situation has occurred on the borehole wing of the repository. Here convergence overcame brine intrusion be­cause of the higher temperatures in the borehole wing and par­ticularly in the HLW field. The sealings of the boreholes have been compacted so strongly by the convergence process that their permeability has become comparable to that of rock salt. About 80 years after the onset of the scenario the boreholes are practically closed and subsequent contact between brine and wastes is impossible. The drifts above the boreholes are about to close at approximately 120 years where we put our flashlights. Although some brine could intrude, it is less than the water content of natural rock salt, so that no rele­vant migration of brine is expected. Reasonable variations of the model parameters did not change this situation.

The situation is different as regards the MLW boreholes. Some of the boreholes see enough brine for corrosion of canis­ters and leaching of radionuclides from the waste matrices. Although convergence is slowed down with the flooding of the boreholes, it still proceeds and squeezes contaminated brine into the overlying drift system and eventually into the cen­tral field where brine intrusion first started. Without the existence of the chamber field the contaminated brine would be pressed out of the repository the same way it came in. How­ever, because ot the lower temperature and the larger initial volume of the chamber field, the chamber field still remains in the phase of brine inflow so that contaminated brine from the MLW boreholes flows into the chamber field together with fresh brine from the main anhydrite until all drifts and cham­bers are filled with brine. A similar cyclic behaviour which reduces the radionuclide release from the repository also exists between the different chamber fields.

32

500 years after the beginning of brine intrusion all voids of the repository are filled with brine. ~n the final phase, starting from then onwards, MLW boreholes and chambers squeeze contaminated brine out of the repository through the main an­hydrite to the aquifers of the overlying strata and from there radionuclides flow into the biosphere with the groundwater. The maximum concentration of radionuclides in the biosphere will show after about 10.000 years.

The coupled equations for brine flo\/, hydraulic pressure, and convergence for the drift network of the repository have been solved with the EMOS computer code at the Technical Uni­versity of Berlin. Convective exchange of brine between neigh­bouring sections of the repository via temperature and density gradients and gas flow are also included. Mobilization of ra­dionuclides is describ~d by models for corrosion of canisters and waste matrices. The transport of radionuclides is con­trolled by the motion of the brine as well as by diffusion, precipitation and dissolution of the radionuclides themselves.

The convective brine exchange mechanisms - including dif­fusion - turned out to be rather unimportant for the radionu­clide release from the repository when compared with the flow forced by cpnvergence. The influence of the primary source term, i.e. the mobilization of radionuclides from the wastes, is discussed in detail in the following.

Mobilization of Radionuclides

Corrosion of the waste containing canisters can start with the small amount of brine migrating from the surrounding rock salt into the deposit locations as soon as the canisters have been emplaced. Later inflow of brine from the drifts sets up a continuous fluid phase apIa to transport radionuclides. Func­tions describing th~ failure of canisters and mobilization of radionuclides are defined from this point onwards.

The maximum canister lifetime depends on the type of ca­nister used for a particular kind of wastes. It takes on the following values:

10 a

40 a

200/300 a

500 a

wastes without fixation anQ HLW canis­ters

concreted wastes

concreted wastes with lost concrete shielding

cast iron canisters

33

The maximum canister lifetime is diminished by the time between deposition of the waste and the first contact with brine flowing in from the drifts. However, by definition it does not become less than 10 % of the original value. A con­stant relative rate of canister failure, which is the inverse of the reduced maximum canister lifetime, is assumed.

As soon as a canister has failed, waste matrix corrosion starts. A constant rate of matrix corrosion, except for vitri­fied HLW, is asssumed. The matrix lifetime depends on the waste form with the following values used:

10 a wastes without fixation

50 a homogeneous cOncreted wastes

100 a heterogeneous concreted wastes

500 a organic matrix

The relation for the HLW corrosion rate per unit surface reads:

j(T) = j(T=473K) j(T=473K) Q R

exp(-Q/R(1/T-l/473K» = 19 m- 2 d- l = 50 kJ mol- l = 8.31"10- 3 kJ K-lmol- l

Q is the activation energy and R is the universal gas con­stant. A canister contains 150 1 of glass. The geometrical surface of 1.7 m2 is increased by a factor of 10 because of fractures in the glass matrix. Via the temperature dependence a time dependence also enters into the corrosion rate. At a constant temperature of T = 473 K, which is the peak tempera­ture the glass matrix reaches, corrosion would be completed after about 80 years. On the other hand, starting with a peak temperature of T = 363 K the time span would be 40.000 years, primarily because of the fact that the later temperature drop (after about 100 years) further decreases the corrosion rate.

As there is a possibility of varying the peak temperature either by interim storage of HLW before final disposal in the repository or by dilution it is interesting to note that a mi­nimal mobilization of HLW can be achieved both at very low temperatures (363 K) and at very high temperatures (473 K). At low temperatures the slow corrosion of the glass matrix is the limiting factor while at high temperatures it is the large convergence rate which leads to a closure of the boreholes be­fore brine can enter. From this we could learn that the optim­ization of the primary source term is not necessarily an opti­mization at the same time for the source term of the reposi­tory as a whole.

34

Variation of Canister Lifetimes

Canister and matrix corrosion functions are generally con­servative estimates based on the assumption that sufficient brine is available for complete alteration of canisters and waste matrices. Table I shows the maximum brine contents for the different types of boreholes and chambers.

Table I

brine volume metall mass cement/glass mass V/m3 M/Mg M/Mg

HLW-B 0 19 36

MLW-B 3 92 242

RS-Ch 206 3700 30000

TL-Ch 34 94 664

TD-Ch 116 1200 9400

Quantities given are per borehole (B) or chamber (Ch)

On the average approximately 1 1 of brine is available per 100 kg of concrete. Corrosion of the metal present will fur­ther reduce the available amount of brine; therefore, the lifetimes given above seem to be extremely conservative. A va­riation of the canister lifetimes , representative also for the matrix lifetimes, has been performed in order to estimate the importance of the primary source term within the scenario. A useful quantity hereby is the cumulative relative release from the repository, which is an integrated release of a ra­dionuclide relative to the initial inventory of the reposito­ry. In Table II the cumulative relative release after 100.000 years (release rates are negligible above this time scale) is given for a selection of important radionuclides among which Tc99 and Np237 are responsible for the main radiation exposure in the biosphere. Besides the reference case variations of the calculation with canister lifetimes increased by factors 10 and 100 are shown. IA129 symbolizes iodine in the form of AgI, while 1129 stands for iodine in more easily solvable form. It can be seen that the cumulative release is not very sensitive to the canister lifetime. This is mainly due to the very slow decrease of the brine contents of the repository in the "squeeze out" phase. Thus it may be concluded that the direc­tion of further investigations should not be towards the re­duction of the corrosion rate but towards a limit to the abso­lute amount of corrosion.

35

That this is possible seems to be guaranteed by the very small ratio of brine to concrete plus metal at the waste locations.

TABLE II

Cumulative relative release from repository after 1*10 5 a

Nuclide canister canister canister linventory lifetime*l lifetime*lQ lifetime*loo M/kg

C14 2.14*10 -3 1. 99*10 -3 1. 22*10- 3 13.3

Se79 2.89*10 -2 2.67*10 -2 1. 84*10 -2 444

Tc99 1. 96*10 -2 1.83*10 -2 1. 38*10- 2 65290

J129 2.93*10- 1 2.59*10-1 1. 49*10-1 260

JA129 1. 92*10 -4 1. 90*10 -4 1. 88*10 -4 15460

Np237 4.45*10 -4 4.36*10

-4 3.72*10 -4

38210

Solubility Limits

A limitation to the absolute amount of mobilization is al­ready set by the solubility limit for a couple of radionu­clides. Table III gives the solubility limits used for differ­ent radionuclides while Table IV show$ the inventories of ra­dionuclides at different waste locations.

Table III

Solubility limits

nuclide Ch + MLW-B HLW-B mo111 mo111

C14 1*10- 3 1*10-1

Se79 1*10-1 1*10-1

Tc99 1*10-1 1*10-1

J129 1*10-1 1*10-1

JA129 5*10- 6

Np237 1*10- 5 1*10-4

36

It is indicated from which location the maximum release of ra­dionuclides originates and where the solubility limit is at­tained. In boreholes the canisters are more densely stacked than in chambers. Therefore, the relative brine contents of boreholes in smaller and solubility limits are more easily met.

MLW-boreholes have the highest inventory of C14 and its solubility limit is attained. In TD chambers no precipitation of C14 takes place and the main release stems from this lo­cation.

Table IV

Radionuc1ide inventories in kg

nuclide total TD-Ch TL-Ch RS-Ch MLW-B HLW-B

r s C14 13.3 0.96 0.0 0.08 12.2 0.0

r Se79 444 1.18 0.0 7.22 43.7 391.4

r Tc99 65290 22.1 0.0 64.2 4270 60930

1129 260 0.49 0.0 r

181 78.5 0.16

lA129 15460 15460 sr 0.0 0.0 0.0 0.0

Np237 38210 29.2 s 365

sr 189 s

37630 0.0

s - meets solubility limit

r - main release

Se79 and Tc99 do not meet their solubility limits. The main release comes from the MLW-boreholes which have the high­est inventories.

The main release of 1129 originates from the RS chambers. The mobilization of iodine in the form of Ag1 (1A129) is strongly suppressed by its low solubility limit.

Np237 attains its solubility limit in boreholes and cham­bers. The main release comes from the chambers because of their higher brine contents.

Present solubility limits are regarded in some cases as very conservative estimates which might be decreased with the increase of understanding of the chemical environments. For radionuclides which attain their solubility limits at all lo­cations, a reduction of the solubility limit reduces the re­lease from each of the locations by about the same factor. Ra­dionuclides which are presently not controlled by solubility need a strong reduction of the solubility limit in some cases in order to affect the source terms appreciably.

37

In Fig. 2 for example the ratio of cumulative release of Tc99 at a given solubility limit to the release at the re­ference solubility limit (0.1 mol/I) is plotted for different locations.

...... Q) L...

(/)

10-2

- 10-3

:E -

10-6 / /

/ /

/ /

/ 0/

/

Tc 99

/ /

/ 0/

/ /

/ /

/ /

····':7t.·~·· ...... . / ....

/ ·0 / \

/ ; / : o :

repository MLW-B RS-Ch

10-5 10-" 10-3

S [molll]

Figure 2: Release of mass M of Tc99 normalized to the re­ference case as a function of the solubility limit S

Dots symbolyze data points while lines interpolate the data points. At the solubility limit 0.1 mol/l the ratio is 1 by definition. Tc99 is controlled nowhere by solubility. MLW boreholes are responsible for the toal release of Tc99 from the repository. As discussed above, a part of the contaminated brine flows from the boreholes into the RS chambers. The net release from there is negative. For solubility limits between 10- 2 and 10-1 mol/l Tc99 is controlled by solubility in the boreholes so that the release is proportional to the solu­bility limit (straight line).

38

with a smaller inflow of contaminated brine into the RS cham­bers the release from there recovers and makes up the main contribution to the total release from the repository at the solubility limit 10-4 mol/l. For solubility limits smaller than 10-5 mol/l, the releases from both of the locations and therefore also the total release from the repository, are pro­portional to the solubility limit. The contribution from the boreholes is an order of magnitude smaller than that from the chambers thus reflecting the ratio of brine flow from the two locations.

We have seen that although Tc99 is not far from being con­trolled by solubility in the reference case, a reduction of 4 additional orders of magnitude is needed until the propor­tionality of release and solubility limit is reached.

Finally, it should be mentioned that many of the results discussed above are also obtained in the case of the reposi­tory containing spent fuel. Particularly the closure of fields bearing the heat producing spent fuel canisters prior to brine intrusion shall be emphasized.

Conclusions

In the scenario discussed above, which is considered to be the most important of the possible scenarios, high level wastes in the case of reprocessed fuel, and spent fuel on the other hand, are prevented from coming into contact with in­flowing brine by the rapid convergence of heated rock salt. Canister and matrix properties of these wastes are therefore irrelevant within this scenario.

It can be concluded that optimization procedures should not be restricted to the primary source term but be applied to the source term of the multibarrier system of the repository as a whole. For the rest of the wastes the release of radio­nuclides from the repository also does not depend sensitively on the properties of the primary source unless the total amount of corrosion of the waste matrices can be limited which is especially relevant for radionuclides which do not attain their solubility limits. The hope to limit corrosion in the forthcoming calculations is based on the fact that only a small amount of brine is available (1 1 brine for 100 kg con­crete). A better understanding of the chemical conditions in saturated brine in contact with wastes and under the influence of radiolysis may also lead to a reduction of solubility li­mits which is mainly important for radionuclides (e.g. C14 and Np237) which are already solubility controlled in the present study.

References

[1] PSE-AbschluBbericht, Fachband Nr. 16, in preparation.

39-40

SOURCE TERM MODELLING IN THE SWEDISH KBS - 3 STUDY

Ivars Neretnieks Department of Chemical Engineering Royal Institute of Technology S - 100 44 Stockholm, Sweden

ABSTRACT

The release rate of the individual nuclides from spent fuel in a deep geo­logic repository in crystalline rock is modelled. After canister breakdown when the fuel is wetted and starts to dissolve, the released nuclides migrate out into and through the backfill. The backfill which consists of a compacted bentonite clay has low permeability and effectively permits transport by diffusion only. The clay has strong sorptive capacity for most nuclides and retards their movement through the barrier. Many of the shorter lived nu­clides Am241, Cs137, Sr90 , Pu240 decay considerably during their passage of the buffer. The longer lived nuclides will eventually be released to the outside of the buffer without significant decay.

During the ensuing semi-stationary phase the release rate will be determined by the dissolution rate of the uranium oxide matrix and the congruent release of the other nuclides. There are some exceptions notably Cs and I which are known to have migrated out of the fuel crystals during the reactor operation.

The dissolution rate of the uranium oxide is determined by its solubility and the various diffusion barriers between the fuel and the flowing water. The~e consist of 'c1ay backfill, possibly the clay which has intruded into the fis­sures and the slowly moving water in the fissures. The transport in the various barriers has been modelled. The dominating resistance in the cases studied was located in the slowly moving water in the narrows fissures.

The situation is somewhat complicated by radio1ysis which may give rise to an oxidizing environment nearest to the fuel. This increases the solubility of many important species (U, Np, Tc •• ) and increases the release rates. The escaping oxidized species and the oxidizing agents (H202, 02) them-selves will move out into the rock where they will react with divalent iron minerals in the rock. At the sharp redox front which develops within a few meters (to maximum a few tens of meters) downstream from the canisters, the redox sensitive elements (U, Np, Tc •• ) precipitate. They will be released to the flowing water outside the redox front in much smaller quantities than that by which they were released from the fuel.

41

INlRODUCTION

The present study is based on the conditions which will prevail in a reposi­tory for radioactive waste in crystalline rock at large depth as in the Swe­dish Nuclear Fuel Safety Project (KBS) concept.

The canisters with spent fuel are emplaced in holes in the floor of tunnels at 500m depth. The bedrock is crystalline e.g. granite or gneiss and has a low hydraulic conductivity Kp <10-10 m/s. The backfill consists of either compacted bentonite in the holes or a mixture of bentonite and quartz sand. The tunnels are backfilled with the latter mixture. See figure 1.

The bentonite, when wet, has a strong swelling pressure and very low hydrau­lic conductivity Kp < 10-12 m/s. In the KBS concept the maximum tempera­ture rise will be to less than 800 C on the canister surface. The copper canisters are designed not to corrode for very long times, they will not be penetrated in less than 100 000 years. If and when a breach occurs, the near field temperature is very near the ambient temperature in the bedrock, 250 c.

The present paper is concerned with the transport of species to and from the degraded canister when all the spent fuel or high level waste is exposed to the groundwater in the bedrock. The paper concentrates on the rate of release of radionuc1ides to the flowing water. It describes a method whereby input data to a far field migration model may be obtained.

CONCEPTUAL MODEL AND MAIN MECHANISMS

The basic approach to the problem has been to determine how much of the water which flows past the repository will exchange species with the canisters. The exchange is limited by the diffusional resitance in the nearly stagnant water in the backfill as well as the diffusional resistance in the slowly moving water in the bedrock. As the canisters are spaced fairly far apart it may be assumed that they do not influence each other.

The basic scenario is the following. At some time the canister material has degraded. The surface of the waste is wetted and the waste reacts "instanta­neously" with the water, dissolving some constituents of the waste up to the solubility limit of that constituent in the chemical environment near the surface of the fuel. Some constituents like iodide and cesium will dissolve entirely in the water, others like silica of the glass, uranium of the spent fuel or nuclides like thorium will attain their equilibrium concentration in the water in contact with the waste.

42

There will be an instationary phase when the species diffuse out into the backfill and into the moving water. This is relatively short in relation to the total leach times of interest for the majority of the important nuclides, but some of the shorter lived nuclides decay considerably. During the ensuing stationary period the species are transported away from the fuel at a rate permitted by the rate of diffusion through the backfill and by the rate at which the flowing water can carry them away. The concentrations of the species in water in contact with the surface of the waste is at the solubi­lity limit of the species or determined by the release rate from the matrix.

It is conceivable that all nuclides will not be released to the water at a rate large enough to keep the concentration at the solubility limit. This may occur if the release is dominated by the dissolution rate of the matrix of the spent fuel (congruent dissolution) where most of the waste (95 %) con­sists of U02. The nuclides in the crystals of the fuel may only be released at the rate at which uranium is dissolved and transported away •. Both the congruent dissolution case and the case limited by individual solu­bilities is treated.

The paper is structured as follows

o Instationary period o F10wrate of water in the near field of the canister o Diffusion into the water flowing outside the backfill o Diffusion in the backfill o The influence of redox processes o Release rates to the far field

TRANSPORT OF DECAYING RADIONUCLIDES FROM SPENT FUEL THROUGH THE BACKFILL -INSTATIONARY PERIOD.

The migration of the radionuc1ides in the backfill takes place in a fairly complex geometry. The nuclides will migrate from the spot where they are dissolved, out into the backfill and then from the backfill into those sections of the rock which are intersected by fissures carrying water. A three dimensional treatment of the transport in the backfill by numerical methods has been done by Andersson et a1. (1982) [1] assuming that all the canister surface has the same reactivity and that fissures intersect the re­pository hole in a regular fashion. The main emphasis of this treatment was on the steady state part of the transport.

Here we want to study that part of the leaching when the concentration pro­file is building up in the backfill. The main matrix constituent i.e. the uranium from spent fuel, is assumed to have reached a steady state earlier than the sorbing nuclides which are retarded during their transport. This is a fair approximation as the uranium under oxidizing conditions will not sorb strongly and thus will not be much retarded in the clay.

43

The sorbing nuclides will be considerably retarded during the instationary phase and if the time for first arrival at the outer boundary of the backfill is very large compared to the ha1f1ife of the nuc1iqe, the nuclide may decay to insignificance before it reaches the outer boundary of the bac~fi11.

A simplified treatment to assess the importance of the retardation is done below. The geometry of the backfill is modelled as a very thick (semi-infinite) slab of backfill. Nuclides ar~ released into one side of the slab by congruent dissolution, and the transport rate of the nuclide past a plane at distance Zo is calculated. As long as the transport rate past Zo is very small, the thick. slab is a fair approximation of a slab with thick­ness Zo with many of the boundary conditions applicable to the subsequent transport in the bedrock. When the radionuc1ides are leached from the spent fuel they will start to mi­grate into the backfill. Some of them like Cs and I will, except for a short initial phase, be released by congruent dissolution. This means that they will be released in proportion to their concentration in the fuel at any given moment. For all practical purposes the decay of the main matrix consti­tuent U238 may be neglected in comparison to all other nuclides. The mass of the shorter lived nuclides will decrease in the f~e1 due to decay accor­ding to m = mo e-)..t if they ha~e no precursor. mo is the mass of nu-clide i at the time of burial. When the matrix di~solve~ it w+11 attain a uranium equilibrium concentration at the surface of the fuel equal to the so­lubility cU,sol of uranium in the ground water. The matrix dissolves away as components diffuse out throu~h the backfill. The disso1¥tion rate can be determined. This has been discussed by Neretnieks (1982) [2] and will be treated later in this paper. The rate for the U438 release from the matrix is NU. Other nuclides are released in proportion to their abundance in the fuel. Nuclide "i" is released with rate Ni

(1)

First those nuclides which have no solubility limitation are treate~. In this case all released nuclides are assumed to move into the backfill.

No solubility limitation.

For brevity we omit the index for nuclide "i". The transport through the backfill is described by equation (2) for the case of a flat backfill barrier. (We neglect the curvature in this analysis.)

2 dc = D ~ - AC dt a dZ 2

(2)

for a decaying species.

44

The initial condition is c - O,z > O,t - 0 The boundary condition for t > 0 at z - 0 is

N = NU mo/mu e- At = Dp€p A ~~ I z=O (3)

which indicates that all released nuclide is transported into the backfill. The uranium dissolution rate NU is assumed to have become stationary and is thus constant. The boundary condition 3 implies that the congruent dissolu­tion goes on forever. Although this obviously is not true, the dissolution time for the matrix is so large (many millions of years) in low permeability crystalline rock (Neretnieks and Rasmuson 1983) [3] that those nuclides which survive so long also will have attained a constant concentration in the back­fill for distances larger than the envisaged backfill thickness.

_ At ~. ~t c=be z----H Z

where b = N.~ /(m __ AD € ) U 0 U--p P

2 z

e a - erfc (z ) - 4D t } 2JDa t

dc and for the gradient we obtain dz to be

dc -= dz

and thus

- At e

erfc y 2 ~t

z erfc 0

2~t

(4)

(5)

(6)

The rate of transport of radionuc1ide "i" past Zo as a fraction of that entering the backfill at z = 0, and t = 0 is

N z

o /N

o (7)

Equations (4) and (7) consists of two parts. The first part accounts for the radioactive decay. The second part describes the change of concentration and flux of a stable species with time and distance. The convenient separability of the solution is due to the presence of the same decay term in equation (7) as well as in the boundary condition at the inlet, equation (3). If the im­portance of the outlet boundary condition is deemphasized, i.e. mainly the early times are considered, we can study the nondecaying part of the solution independently of the decaying part. The treatment above has considered an infinitely thick backfill barrier and linear transport of the diffusing species. The backfill in a repository will have finite length and cylindrical shape more than slab shape. It has been shown (Neretnieks 1982) [4] that for "short times" and a reasonable ratio of the outer to inner radius rout/rin <2 the approximations above give small errors and the simple solutions above for the infinite plane case give good results.

45

For screening purposes and for first estimates it is deemed to be sufficient to use the solutions for any of the geometries and boundary conditions. For cz/co< 0.05 the results differ very little.

Solubility limitation

In some circumstances some nuclides e.g. plutonium and thorium may have so low solubility that they precipitate at the fuel surface as the matrix dis­solves away. The boundary condition at z=O then becomes

c = Co for t > 0 at z = 0 (8)

The following solution is obtained by Laplace transformation (Neretnieks 1984) [5]

c c = ...2.

2 { -z~ A

e Da zW=

er f c (-::-2 -\f15 ...... z-a=t - VM:) + e a

+ 0 t )} (9)

Equation 9 may also be used to obtain the flux past a plane at distance zOo As this flux cannot as easily be related to the release rate as for the case with no solubility limitations another approach is chosen. The concentration Cz at distance Zo is related to the solubility concentration Co instead.

o

A generalized approach

For the case with no solubility limitation Equation 7 can be transformed to

N IN z 0 o

-T = e H

erfc 2 J[

Equation 10 has only one parameter H I A

where H = Zo \,%

(10)

(11)

The maximum of equation 10 is then a unique function of H. Figure 3 shows this function. The figure also shows at which times the maximum occurs.

This type of curve can be used to get an indication of which nuclides will strongly and which will be weakly affected by the presence of the buffer.

For the case where there is solubility limitation equation 9 can be written

(12)

clc =! { e-H erfc( H o 2 2 yT

if the same variable transformation is used as previously.

46

As in the previous case equation (12) is a unique function of H. It has no maximum but has an asymptotic value.

clc o lim

= e -H

T + 00 (13)

Figure 3 shows this curve also. It is seen that for H-va1ues larger than about 30, clco becomes very small < 10-15 in the same way as the maximum release N declines.

Some calculated results

Figure 2 shows the release NINo for Pu239 for a case when there are no solubility limitations. The buffer thickness is taken to be 0.375 m. The apparent diffusivities in table 1 are used for the calculations. The figure also shows the release in a case where there is no retardation in the buffer. The peak release (5.4. % of original maximum leach rate) takes place after 40'000 years. At this time decay accounts for a decrease by a factor of 5.3. and the retardation effect contributes with another factor 3.5. The latter is somewhat on the low side to assure the applicability of equation (10). It somewhat underestimates the peak height and the portion on the downhill side.

Table 2 shows values of H for some nuclides of interest. For all nuclides the maximum release NINo as well as the maxium (asymptotic) concentration clco are given in the table.

The only nuclide of those considered which decays to insignificance is Am241. The nuclides Am243, Pu239 , and Pu240 will be noticab1y retar­ded. Sr90 and Cs137 will be somewhat retarded. Th229 comes predominan­tly from the Np237 chain which passes the backfill barrier and will thus not be influenced by the backfill.

STATIONARY PERIOD

Water flow around the canister

The water flow in the vicinty of the canister will be somewhat larger than in the undisturbed rock far from the caverns where the rock is not influenced by mining activties. Assuming that the bedrock can be treated as a porous medium and that the rock is much more permeable near the canister than far away, it can be shown that the flux through an infinitely long cylindrical hole is twice that in the undisturbed rock (Tietjens 1960) [6]. The same applies to flow in a fissure intersecting a hole perpendicular to its axis.The presence of more fissured rock near the backfilled holes acts as an empty hole in the above description. The presence of an impermeable area - the backfill - in­side the permeable area does not influence the flow pattern in the rock. There are no detailed data on the size and orientation of the fissures in the rock. In the following the f10wrate in the rock surrounding the backfill is taken to be twice that in the far field rock.

47

The f10wrate of water in the bedrock in the far field can be determined if the hydraulic gradient and conductivity is known, the f10wrate is uo·Kp i and the velocity of the water is up·2uo/ER in the near field.

Mass transfer to the water flowing past the backfill

Two cases are treated. In the first case we assume that the bedrock around the backfill can be treated as a porous medium surrounding an infinitely long impermeable medium (the backfill). The assumption of an impermeable backfill is a good approximation for the bentonite clay whiCh has a very low hydraulic conductivity compared to the bedrock. The transport by diffusion of dissolved species is much faster than by flow (Neretnieks 1977) [7]. Transport by flow in the backfill is therefore neglec­ted.

In the second case we assume that the bedrock consists of individual fissures intersecting an infinitely long clay cylinder at regular intervals. All water flow takes place in the fissures. The mass transfer to the flowing water is treated as follows: The water which approaChes the backfill has a known con~ centration Coo of the species of interest. As the water approaches and flows past the backfill it picks up (or is depleted of) the species whiCh diffuse in the water from (or to) the outer surface of the backfill. See Figure 4. in the following only outward transport will be described. Inward transport may be described in the same way. During its passage past the backfill the flow­ing water will pick up more and more of the diffusing species. When the water ~eaves the vicinity of the backfill, it has taken up an amount of species whiCh is mainly dependent on the residence time of the water at the surface of the backfill and the diffusivity of the species. A rigorous treatment of this case is given by Andersson et a1 (1982) [1]. A simplified approaCh was taken by Neretnieks (1978,1980) [8,9]. Assuming that the penetration depth of the diffusing species is short com­pared to flow length, the mass N transferred during this time can be obtained by solving the diffusion equation for the appropriate initial and boundary conditions. The solution is given in detail by Bird et a1 (1960 p. 537) [10]. The results can be writte~n

~D

N = 27TLr 3 Ea (coo - c3) : t = ~ (coo - c3) (14) res 3

Chambr~ and Pigford (1983) [11] arrive at the same result but a constant 2/~ higher (i.e. 13% higher release rate) by a slightly different method. They also give a range of validity to the use of eauation (14). It is valid for

u r3/D < 4 (15) p w

48

Mass transfer in the backfill

The distinction between the case where the rock is a "porous" medium and the case where it is a sparsely fissured rock becomes important when the migra­tion in the backfill is to be treated. In the first case the diffusing spe­cies can enter the water in the rock everywhere along the backfill/rock in­terface, whereas in the second case the species must find the widely spaced individual fissures.

The diffusional transport through a cylindrical barrier can be written (Bird et a1 1960, p. 287) [10].

(16)

If backfill material can be injected into the porous bedrock to some distance r3 (or into the individual fissures) there will be an additional transport resistance. This can be written ana1oguous to equation 16.

(17)

~2is the effective diffusivity in the backfill injected into the pores of the bedrock. As it is imp1icity assumed (not quite true) that c1, c2 and c3 are con­stant around the cylindrical circumferences, c2 and c3 can be eliminated from equations 19, 20 and 21 give

(c1 Coo )

N = 27TL S 1n(r3/r2) (18) 1n(r2/r1) 1 -+ + j4D D1 ° D2 ER r3 ERV w

7Ttres

R1 R2 R3

o/s is the equivalent fraction of the clay that is effectively utilized by the diffusing species.It is explained below. The three terms in the denominator are proportional to the resistance to mass transfer in each section R1, R2 and R3 respectively. They can be used to compare the importance of the various diffusional barriers.

49

RoCk is a fissured medium

A case where the roCk is fissured with equally spaced horizontal fissures is also modelled. The diffusing species from the degraded canister will have an effectively de­creasing crossectional area to migrate in until it reaches the fissures. See figure 5. Assuming that all transport takes place between a circle segment 2b wide and an outer segment S wide at a radial distance S - 2b, an approximate average width for mass transport is obtained from the expression for diffu­sion between two concentric cylinders.

= ,.;;;.S_-....".;;;2,.;;;.b S

In 2b =

2b « S

S S

1n 2b

(19)

This expression can only be expected to be reasonably valid when the half distance between the fissures S/2 and the thic~ness of the buffer are nearly equal. As will be shown below the resistance in the buffer is small compared to other resistances and the inaccuracies introduced by the approximations will have negligible influence on the final results. Rigorous methods are available however and have been tested (Andersson et a1 1982) [1]. Thus in the case of fissured roCk only a part of the backfill length= O/S is effectively utilized for mass .transport. The resistance in the backfill RI then increases by a factor S/ 0 as com­pared to the case where the rock is a "porous" material. For a porous me­dium 0 = S. Andersson et a1 (1982) [1] have shown that the approximations leading to equation 19 do not introduce grave errors. The transport rate N of a species may also be expressed as an equivalent water flow rate Qeq which would increase in concentration from c1 to coo •

Q = N/(c - c ) eq 1 . 00 (20)

Equation (18) can be used to determine the release rate of a species when the concentration at the fuel surface (c1) and in the waters far away (coo) is known.

Since the diffusivities of the nuclides in water do not differ appreciably from each other and are of the same order of magnitude as for other small dissolved ions, they will be transported equally fast with the same driving force (concentration difference). This is the difference between the concen­tration at the fuel surface Co and the concentration in the water far away, where it is O. The nuclide transport can therefore be written:

(21)

where Ni is the quantity of nuclide "i" that is transported at a concentration difference of coi - O. The parameter R is equal to the sum of the transport resis tances in clay and in fractures. R is determined with the aid of the model described above and is little dependent on nuclide type.

50

Equation (21) can also be interpreted in the following manner: Qeq is the equivalent water flow, that arrives at the canister with the concentration 0 and leaves it with the concentration cio'

N = Q c. (22) i eq 01

Qeq has been calculated for the 5 m long canister. Table 3 gives values of Qeq at different water f10wrates Uo in the rock. Tao1e 4 shows that the resistance in the flowing water in the fissure is 10 times larger than that in the backfill and thus totally dominates the release rate.

The outward transport of uranium will determine the dissolution rate for the uranium dioxide matrix. The concentration of uranium at the surface of the pellets is determined in part by its solubility in the waters in question and in part by whether the dissolution rate is large enough to supply the water at the surface of the pellet with new uranium at the same rate as it diffuses away. Since no reliable data are available on the dissolution rate of spent fuel, it is assumed to be rapid, i.e. it is not limiting. The concentration at the surface will then be equal to the solubility of the uranium. This is determined primarily by the water's carbonate content and pH and by the wa­ter's redox potential Eh. The redox potential in surrounding water in the rock is negative, but can be positive at the surface of the pellet due to radio1ysis. At positive Eh, the solubility of the uranium is determined by the concentration of carbonate ions. At the highest carbonate contents HC03 = 275 mg/1 in Swedish groundwaters in crystalline rock, the solubility of the uranium = 360 mg/l. With Qeq = 0.57 l/canister and year, and with CoU = 360 mg/1, equation (22) gives out

NU = 0.57 * 360 = 205 mg U/year and canister.

Each canister contains 1.4 tonnes of uranium, which means that it takes about 7 million years to dissolve the canister's uranium content. This also pre­sumes that the production of oxidizing species via radiolysis is sufficient to oxidize the uranium at at least the same rate as it is carried away.

The nuclides contained in the uranium dioxide crystals are liberated as the crystals dissolve. They will have a concentration on the surface of the pel­lets that is proportional to their concentration in the pellets. Thus, nu­clide "i" will have the concentration:

c . 01 c . 1 X. uran1um, so 1 (23)

where Xi = the fraction of nuclide "i" in the pellet and c is the solubili­ty of uranium. The nuclide "i" is transported out to the water in the same manner as the uranium. The flow is as before Ni = Qeq coi. If the nu-clide "i" can dissolve in the water in the concentration cio' it will be transported out at the same relative rate as the uranium.

51

Some nuclides, e.g. Th and Pu, have solubilities that are lower than cio' They therefore precipitate on the surface of the pellets and are only trans­ported away to a small extent. This is determined by the solubility of the individual nuclide. Thus, if the solubility ci'sol is smaller than cio according to (23), the outward transport of nuclide "i It will be determined by:

N = c Q i i,sol eq (24)

The outward transport of plutonium and thorium is greatly restricted by their low solubility.

Some cesium and iodine has been enriched at the surface of the pellets during reactor operation. This fraction of the cesium and iodine is assumed to be immediately available for leaching. Cs137 will decay during the life of the canister. Cs135 and 1129 reach the water outside virtually unaffected by the buffer. However, since the canisters ~re penetrated at varying times and spread out over a very long time span, the accessible portion of these nu­clides will, on the average for the repository on the whole, be released at a rate that is determined by the rate of canister penetrations. The majority of these nuclides as well will be released when the uranium dioxide dissolves. Since the dispersion in time of the disintegration of the canisters is com­parable to the leach time for the uranium, the releases of these nuclides do not differ appreciably from those of the others.

Radiolysis and its consequencies

The ionizing radiation can split the water into hydrogen and oxidizing com­pounds such as oxygen and hydrogen peroxide. The radiation can penetrate the canister wall and other solid materials and act at a distance of several de­cimetres from the canister. It is however, greatly attenuated on its passage through the solid materials, and its intensity on the outside of the canister is very weak (Christensen and Bjergbakke 1982) [12]. The a and S radiation has a very short range in solid materials and cannot have any effect until the water has come into direct contact with the uranium dioxide matrix. The effective range of the a radiation in water is about 0.03 mm. When the canis­ter has been penetrated and water has got in to the surface of the pellets, a radio1ysis can take place. Calculations have been carried out by Christensen and Bjergbakke (1982) [13]. They have assumed that the entire surface of the uranium dioxide pellets is covered by a layer, 0.03 mm thick, in which the energy from the a radiation is spent. In each canister, approx 4.2 1 of free water could then be subjected to a radio1ysis. This is greatly exaggerated, since the corrosion products from the can~ster material and the uranium di­oxide occupy a larger volume than the original materials, so that even if this volume exists in the gap between the uranium dioxide pellets and the zirca10y cladding when the ca~ister is intact, it is greatly reduced after a corrosion attack. Table 5 shows the radio1ysis at different points in time expressed as equivalent generated quantity of hydrogen peroxide and hydrogen. These quantities of radio1ysis products are probably greatly exaggerated. Radio1ysis on this scale has not been observed at Oklo (Curtis and Gancarz 1983) [14], where a natural reactor has been active for

52

some 300 000 years. Radio1ysis there has been 100-500 times lower, figured for a comparable quantity of fuel and burnup, than the theoretically maximum figures. Even low levels of iron dissolved in water catalyze the reverse re­action between hydrogen perox~de and hydrogen. This can reduce the net pro­duction of hydrogen and hydrogen peroxide by a factor of several hundred (Christensen and Bjerbakke 1982) [13]. The consequences of maximum radio1ysis are explored below. However, the radio1ysis is expected to be no more than 1/100th of the reported maximum values due to limitations in water supply caused by expansion of the corrosion products, hydrogen outflow and catalyzed reverse reaction.

The hydrogen peroxide produced by radio1ysis is a powerful oxidant which can attack both the copper canister material and the lead and zirca10y that surround the uranium oxide pellets as well as the uranium oxide itself. In the presence of certain substances, hydrogen peroxide can spontaneously de­compose to oxygen and water. The oxygen will thereby act as an oxidant in the same manner as the hydrogen peroxide. The fuel's metal oxides can also be oxidized to higher valence states. The uranium, which is tetravalent in the fuel, can be oxidized to hexavalent. Other nuclides can also be oxidized, the most important being plutonium, neptunium and technetium. Uranium, neptunium and technetium become many times more soluble in.the groundwater at the higher oxidation state. When the uranium dioxide in which the other nuclides are contained is oxidized, these nuclides can be released. They can then be transported out from the repository.

Calculations have been carried out on the effect of the radio1ysis (Grenthe et a1. 1983) [15], where the simultaneous outward transport of radionuclides to the mobile water outside the canister has also been taken into account. The calculations are based on the assumption that chemical equilibrium always exists in a given volume of water, which is determined by the water content of the clay. The hydrogen peroxide formed according to table 5 and water with its content of carbonate and other dissolved species enters this volume. The same water flow carries away the reaction products. The water flow Qeq is determined as before by the geometry of the near field.

The outward transport is determined in part by how much of the radionuc1ides have had time to be oxidized to the more easily soluble forms and in part by the supply of carbonates in the groundwater. The latter constitute comp1exing agents for uranium and neptunium. Calculations show that the copper in the canister - if it is accessible - w.i11 be preferably oxidized and prevent the fuel from being affected almost completely. Since the growth of the corrosion products makes it improbable that the copper will always be accessible for oxidation, the calculations were also carried out under the assumption that the copper does not react at all. In this case, the uranium dioxide is oxi­dized first. Table 6 illustrates this process and how much uranium, pluto­nium, neptunium and thorium has been transported out at different points in time. The total carbonate content has been set at 122 mg/1 C2mmo1/1) in the calculations. The large change in the quantity of Np and Pu dissolved in the time interval 960-9 960 years stems from the fact that the strongest reduc­tant U02Cs) has been consumed after approximately 2 500 years. These

53

calculations show what happens for the case of initial canister damage. A ca­nister that has been penetrated after 100 000 years and where no hydrogen peroxide leaves the system either, will have all its uranium dioxide oxidized to hexavalent after an additional 250 000 years, if the maximum possible radio1ysis is assumed to prevail. If the radio1ysis is limited to 1/100th of this, only about 50 kg of uranium will be oxidized in one million years. That will also be the amount of fuel dissolved in this period. The oxidation times calculated above are exaggeratedly short due to the fact that no lead, zir­conium or copper is assumed to react and that the surface of the pellets is assumed to be covered with a uniform layer of water. The canister materials will be oxidized to a not insignificant extent. The oxidation of uranium di­oxide will then be reduced accordingly. The corrosion products from both the fuel and the canister materials have all larger volume than the original ma­terials. They will therefore fill the original volume that had been available for water. The porosity of the corrosion products must be small, since the canister and the surrounding clay and rock exert a powerful restraint on ex­pansion. These effects should limit the effects of radio1ysis greatly, which is supported by the conditions in the Oklo reactor where, after 1.8 billion years, only a very small fraction of the uranium dioxide has been dissolved (Curtis and Gancarz 1983) [14].

Only a very small portion of the oxidized hexavalent uranium leaves the ca­nister, see table 6. By far most of important of the released nuclides are cesium, strontium, iodine and technetium. These have high solubility and can be transported away at the rate at which they are released from the fuel.

The radio1ysis products can at most oxidize the uranium at approximately the same rate as that at which the water flow in the rock is able to transport it away from the canister. This outward transport in the equivalent f10wrate Qeq is about 200 kg/106 years, provided that enough uranium exists in the hexavalent form, while the generation of hexavalent uranium is about 50 kg during the first million years. The release of the aforementioned nuclides is assumed in the further calculations to take place at the rate at which the uranium can be transported out.

Of the actinides, thorium and plutonium have such low solubility under the conditions that prevail in the vicinity of the fuel that they precipitate in the form of solid phases and are only transported away from the canister to a very small extent.

Some of the hydrogen peroxide and other oxidation products, such as hexava­lent uranium, may leave the canister and reach the clay and the rock outside the clay. Both the clay and the rock contain bivalent iron compounds, which are readily oxidized to trivalent iron (Torstenfe1t et a1. 1983) [16]. When all available divalent iron in a zone has been oxidized, conditions in the water become oxidizing and many actinides become more soluble than under the reducing conditions that normally prevail in rock where there is still diva­lent iron. Thus, an oxidizing zone can be formed nearest the canister, which slowly expands as more oxidizing substances are transported out.

54

Calculations of the propagation of this so-called "redox front" have been carried out (Neretnieks 1983 [17], Neretnieks and Kslund 1983 a,b {18,19]). In these calculations, it is assumed that all produced hydrogen peroxide or an equivalent quantity of other oxidizing species penetrate out to the clay and rock. When all iron in the clay has been consumed, the hydrogen peroxide diffuses into the rock and reacts with the iron there. In fractured rock, the hydrogen peroxide is also carried away with the flowing water in the frac­tures, but diffuses from there out into the microfissures in the rock and oxidizes a rock volume downstream of the canister. Figure 6 illustrates this propagation.

The propagation of the front downstream of a canister can reach a maximum of some 50 metres in one million years in relatively iron poor rock (0.2% Fe (II» (Neretnieks and Aslund 1983 a) [18] if the greatest possible radiolysis is assumed to take place and all hydrogen peroxide goes out into the rock. Swedish crystalline rock contains 1-10% divalent iron (Torstenfelt et al 1983) [16]

In practice, the front only moves a few metres. The transverse propagation of the front by diffusion has one negative consequence Radionuclides that are readily soluble under oxidizing conditions spread out in the oxidized region. When these radionuclides come to the redox front and precipitate, the water in a much larger cross-sectional area of the rock will be saturated with the radionuclide than the very limited cross-sectional area in the vicinity of the canister. If the redox fronts from many canisters reach each other, most of the water that passes the total cross-sectional area of the repository will be saturated with the nuclides to the concentration to which they can dissolve under reducing conditions. Figure 7 illustrates the propagation of the radiolysis front sideways in a fracture and how a radionuclide that is highly soluble under oxidizing conditions precipitates to the lower solubili­ty at the redox front and is transported further at the maximum concentration csol. Under unfavourable circumstances, all water downstream of the reposi­tory can become saturated to this concentration. Since, on the average, 150 m2 of the repository's cross-sectional area can be ascribed to each canis­ter, the water flow per canister can be

[1/(m2 year)] (= 15 l/year (25)

in central case)

Since the radiolysis will in practice be much less than the maximum possible, the transverse propagation of the radiolysis front is far less. Calculations show that only a small portion of the water downstream of the repository will become saturated in the manner previously described. In the further calcula­tions, however, it is assumed that all water will be saturated.

55

RELEASE OF NUCLIDES TO THE FAR FIELD

A radionuc1ide with high solubility under oxidizing conditions precipitates at the redox front. The water downstream of the redox front has a low concen­tration cso1, but the front has been broadened and a larger water flow is transporting the nuclide.

In the case of those nuclides that are not affected by the redox front, lea­ching takes place with a flow of:

(26)

and

Ni = Qeq Ci,sol (27)

and the two expressions distinguish the case where the 9utward transport is not limited (26) from the case where the outward flow is limited by the solu­bility of the nuclide (27).

In the case of those nuclides that precipitate at the redox front, the flow beyond the redox front is:

N = Qred cred i eq i,sol

(28)

Thus, the release to the far field can be determined by:

1) The outward transport rate of the uranium oxide; equation (26) (or pos­sibly the oxidation rate),

2) The fact that some nuclides have low solubility and do not dissolve when the fuel (U02) is transported away or altered; equation (27); the release of these nuclides may be less than by mechanism 1),

3) Those nuclides that precipitate at the redox front - equation (28) -will be released at a lower rate than that due to mechanisms 1) or 2). For species e.g. uranium with more than one isotope the different iso­topes are re1ased in relation to their concentration in the precipi­tate. The sum of them making up the solubility concentration.

Trivalent and tetravalent actinides at low concentrations will probably be coprecipitated together with uranium when uranium precipitates at the redox front. This has been shown experimentally for plutonium trenthe et a1 1983) [15]. An equally large fraction of the actinide precipitates as of the main component, uranium. The following actinides can be expected to be coprecipi­tated: Pu, Np, Th, Am. They will be transported to the far field in quanti­ties reported in the righthand column (7) in table 7.

56

With this precipitation mechanism, the plutonium isotopes, neptunium and tho­rium would deliver a nuclide flow to the far field that is 30-1000 times lower than their own solubility permits.

Since this mechanism has as yet only been shown experimentally to apply for plutonium, it is not used in the safety analysis.

The release of the various nuclides is shown in figure 8 for a case where the canister has degreded after 105 years.

DISCUSSION AND CONCLUSION

The models discussed here are based on some rather straightforward concepts: diffusion and retardation by linear reversible sorption in the backfill, dif­fusion in slowly moving water in fissures and reactions of oxidizing species with reducing minerals in the rock.

Various simplifying assumption have been used to make the mathematical model­ling so simple that reasonably simple analytical and approximate solutions have been possible to derive. Most of the simplified descriptions have been compared to more rigorous calculations using numerical techniques to handle the complex geometries involved e.g. 3D modelling of the migration in the backfill, 2D advection dispersion + 1D matrix diffusion and reaction to de­scribe the redox front migration etc. It was found that the simple analytical solutions perform surprisingly well in most cases. Even in the considerably more complex numerical solutions many important simp1ifiying assumptions have had to be made. Furthermore the data presently available on fissure geometry, width, spacing etc. are far from exact. In many cases the poor data and the inaccuracies introduced by various assumptions seem to introduce more varia­bility than even rather strong simplifications in the models. The simple mo­dels helped to understand many of the involved processes better than the "rigourous" solutions. The latter were however necessary to use also in order to understand the limitations and errors introduced by the various simplifi­cations. At present parts of these models are being further developed.

Acknowledgements This work has been done within the Swedish nuclear fuel safety study KBS-3 [20] for the Swedish nuclear fuel supply company.

57

REFERENCES

/1/ Andersson, G., A. Rasmuson and I. Neretnieks: Migration model for the near field. KBS TR 82-24*, November 1982.

/2/ Neretnieks, I.: Leach rates of high level waste and spent fuel-limiting rates as determined by backfill and bedrock conditions. Paper presented at the 5th International Symposium on the Scientific Basis for Nuclear Waste Management. Berlin June 1982, Proceedings North-Holland, 1982, 559-568.

/3/ Neretnieks, I. and A. Rasmuson: An integrated approach to the descrip­tion of radionuclide release and transport in the geosphere. 7th Inter­national Symposium on the Scientific Basis for Nuclear Waste Management. Boston Nov. 1983. In print.

/4/ Neretnieks, I.: Diffusivities of some dissolved constituents in com­pacted wet bentonite clay-MX80 and the impact on radionuclide migration in the buffer. Royal Institute of Technology, KBS TR 82-27, October 1982.

/5/ Neretnieks, I.: Diffusivities of some constituents in compacted wet ben­tonite clay and the impact on radionuclide migration in the buffer. March 1984. Sent to Nuclear Technology.

/6/ Tietjens, 0.: Stromungslehre. Springer 1960.

/7/ Neretnieks, I.: Retardation of escaping nuclides from a final repository. Royal Institute of Technology. KBS TR-30, September 1977.

/8/ Neretnieks, I.: Transport of oxidants and radionuclides through a clay barrier. KBS TR-79 , 1978.

/9/ Neretnieks, I.: Transport mechanisms and rates of transport of radio­nuclides in the geosphere as related to the Swedish KBS concept. IAEA publ. SM-243/108 IAEA Vienna, 1980.

/10/ Bird, R. B., W. F. Stewart and E. N. Lightfoot: Transport Phenomena. John Wiley N. Y. 1960.

/11/ Chambre, P. L. and T. H. Pigford: Prediction of waste performance in a geologic repository. Scientific basis for nuclear waste management VII, 985, Boston, November 1983, North-Holland.

/12/ Christensen, H. and E. Bjergbakke: Radiolysis of groundwater from HLW stored in copper canisters. Studsvik Energiteknik AB, KBS TR 82-02, June 1982.

/13/ Christensen, H. and E. Bjergbakke: Radiolysis of groundwater from spent fuel. Studsvik Energiteknik AB, KBS TR 82-18, November 1982b.

~=--* KBS - technical reports can be obtained from: INIS CLEARING HOUSE International Atomic Energy Agency P.O. Box 100 A-1400 Vienna, Austria

58

/14/ Curtis, D. and A. Gancarz: Radiolysis in nature: Evidence from the Oklo Natural Reactors. KBS TR 83-10, February 1983.

/15/ Grenthe, I., I. PUigdomenech and J. Bruno: The possible effects of alfa and beta radiolysis on the matrix dissolution of spent nuclear fuel. Royal Institute of Technology, KBS TR 83-02, January 1983.

/16/ Torstenfelt, B., B. Allard and T. Ittner: Iron content and reducing capacity of granite and bentonite. Chalmers University of Technology, KBS TR 83-36, April 1983.

/17/ Neretnieks, I.: The movement of a redox front downstream from a reposi­tory for nuclear waste. Nuclear Technology 62 1983 p. 110.

/18/ Neretnieks, I. and B. Aslund: The movement of radionuclides past a redox front. Royal Institute of Technology, KBS TR 83-66, May 1983a.

/19/ Neretnieks, I. and B. Aslund: Two dimensional movement of a redox front downstream from a repository for nuclear waste. Royal Institute of Technology. KBS TR 83-68, May 1983b.

/20/ KBS-3. Final storage of spent nuclear fuel. KBS-3, Report by Swedish nuclear fuel supply company, May 1983.

/21/ Allard B.: Solubilities of actinides in neutral or basic solutions. Pro­ceedings Actinides -81. Asilomar September 10-15, 1981.

59

Notation

DpE:p Ow D1

D2

H Kp L m mU mo N NU Qeq

r r1 r2 r3 R R1 R2 R3

S t

tres

Cross section area half fissure width concentration initial or boundary concentration concentration at fuel or canister surface concentration at clay/rock interface concentration clay/water interface concentration in water far away apparent diffusivity includes sorption effects effective diffusivity diffusivity in water effective diffusivity of backfill clay surrounding the canister effective diffusivity of clay injected in to fissures equation 11 Zo ( A /Da) 1/2 hydraulic conductivity canis ter length amount of nuclide amount of uranium amount initially (at start of leaching) f10wrate of nuclide f10wrate of uranium equivalent f10wrate to carry dissolved species in a concentration equal to that at the fuel surface radial dis tance radius of canister (or fuel)

repository hole clay (in fissures) interface

mass transfer resistance total " of clay backfill

of clay in fissure

or in porous rock fissure spacing time

of flowing water in fissure

residence time of water as it flows around (past) canister ha1f1ife of nuclide distance thickness of barrier

60

m2 m mo1/m3 mo1/m3 mo1/m3 mo1/m3 mo1/m3 mo1/m3 m2/s

m2/s m2/s m2/s

m2/s

m/s m mol mol mol molls molls m3/s(l/year)

m m m m s/m3 s/m3 s/m3 s/m3

m s

s s m m

Greek letters 6 equivalent width for diffusion in clay

backfill Ep porosity of clay backfill ER flow porosity of rock ER=2b/S A decay constant TAt dimensionless time

Subscripts i sol

Compound

nuclide "i" at solubility

1- 9 Tc04 53 Cs+ 8 Sr2+ 25 Ra2+ 25 Pa 1 Th1) 10-2 U1) 1 Np1) 0.4 Pu1) 3-10-2 Am1) 1.5-10-2

m

1) The dissolved species are probably carbonate, hydroxyl and mixed com-plexes, see Allard (Asilomar Sept 10-15, 1981) [21].

Table 1. Upper limits of apparent diffusivity values of some nuclides for breakthrough calculations.

61

Nuclide Tl/2 D -1012 a H /No C/Cso1

max max

1-129 2-107 9 4.15-103 0.97 1 Tc-99 2-105 53 1. 70-10-2 0.92 0.99 Cs-137 30 8 3.59 0.010 0.027 Sr-90 28 25 2.10 0.053 0.12 Ra-226 1.6 -103 25 0.278 0.54 0.75 Th-229 7.3 -103 10-2 6.50 2.9-10-4 1.5-10-3

Pa-231 3.6 -104 1 0.293 0.53 0.75 U-234 2.5-105 1 0.111 0.73 0.89 U-235 7.1-108 1 2.09-10-3 0.98 1 U-238 4.4-109 1 8.38-10-4 1 1 Np-237 2-106 0.4 6.22-10-2 0.81 0.94 Pu-239 2.4-104 3 -10-2 2.07 0.054 0.13 Pu-240 6.6-103 3 -10-2 3.95 6.6-10-3 0.019 Am-241 458 1.5-10-2 21.2 <10-9 <10-9

Am-243 7.4-103 1.5-10-2 5.28 1.6-10-3 5.1-10-3

Table 2. Half life T1/2, Apparent diffusivity and parameter H in a buffer of thickness is 0.375 m. Maximum N/No for soluble species and maximum C/Cso1ubi1ity for solubility limited species.

Uo Qeq 1/(m2 year) l/year

0.01 0.19 0.03 0.32 0.1 0.57 central case 0.3 0.94 1 1.57 3 2.41

Table 3. Qeq at different water flows Uo in the rock for a 5 m long canister.

62

Uo 2b 1/m2y mm

0.1 0.1

S m

1

Rl/Rl R2/Rl R3/Rl

1 o 10

Qeq l/y

0.57

Table 4. Determined equivalent water flowrate Qeq for leaching of spent fuel and relative resistances to mass transfer of the various barriers. Rl/Rl, R2/Rl and R3/Rl are relative resistances in backfill around the canister.

Time after H202 and H2 H202 and H2 discharge from production produced in reactor (mol/y) the time interval (years) (mol)

100 4.6 300 3.2 780 600 1.8 750

1 000 1.31 620 104 0.34 7 400 105 0.025 16 200 106 0.009 15 800

Total produced during 106 years 41 550

Table 5. Theoretical maximum hydrogen and hydrogen peroxide production from radiolysis in 1.4 tonnes of uranium. The radiation is spent in 4.2 1 of water spread out in a 0.03 mm layer on the uranium dioxide surface.

63

Years after Dissolved guantitl, 82 in the time interval deposition U Np Pu

0-1 0.12 2.4_10-11 10-12 1-260 24 7.5-10-9 2.4-10-10

260-960 95 2.1-10-8 7.2-10-10 960-9 960 286 2 000 10-3

9 960-105 104 2 500 0.012 105-106 1.1-105 0.024

Table 6. Dissolution of nuclides from a canister under the influence of radio1ysis as per table 5, Qeg = 1 l/year and the canister ma­terials are inert. The total caroonate content is 2 mM.

Column 1 2 3 4 5 6

Nuci ide Quantity per Congruent dis- Solubility Solubility To far

canister solution limit under limit at field. l1in.

105-year-old (6.8xI06 years oxidizing redox front of col. 3,

fuel dissolution conditions 4, 5 as

(time) activity

g Bq g/y g/y g/y Bq/y

238U 1.4x106

1.66x10 10 0.21 0.21 !:2~~ -4

1.8 234U a) 9.8x10

10 10.5

235 U a) 1. 24xl09 0.13 236U a) l.92xlO 10 2. I 239 Pu 386 8.81x10 11 5.7x10 -5 1.41x10-6 b) 3 200 242pu 862 1.24x10 11

I. 3x 10 -4 3~ 15x~-6 b) 460 237 Np 2 960 7.77xI0 10 '4 -4 4.3x10 1. 2xl0 3 100 230Th 86 6.22xlO 10 1. 3x 10 -5 2.3x10 -7

170 241 Am 4.19x10 -5 5.18xl06 6.2xIO- 12

0.76 ZI,3 Am 1.91x10 -2

I. 45xlOS -----9 2.9x10 21

99Tc 895 5.70x1O ll ---4 I. 3xlO 3x10-6 1 900

129} c) I. 86xl09 270 135Cs c) 1.97xl010 2 900

7

To far

field. co-

precipi ta-

tion with

uranium

Bq/y

94

4.9

3.3

1.8

5.6x10 -4

0.016

Table 7. Releases of certain radionuc1ides to the far field. Based on 1.4 tonnes uranium, Qeq = 0.57 l/y, Q~~d = 15 l/y.

Underlined values are limiting.

a) Follows uranium-238 b) Sum of Pu-239 and Pu-242 is 4.56 • 10-6 g/y c) No solubility limit = Same as in column 6

64

Figure 3

Figure 4

~ (Max t~) I~ (Max(~;o))

o -1

-2

-3

-4

-5

-6

-7

-8

-9

-10

-11

-12

-13

-14

l' at max for N -curve

14

12

10

8

6

4

2

-15 ....j......--l---+--__ I-=~_+--t_....l..-_I_ ....

-3 -2 -1 o 2 log H

Maximum concentration C/C and release N/N versus parameter H. o 0

When the water first comes into contact with the clay a) from which the nuclide diffuses, the nuclide reaches a short distance out into the water. During the progressively longer contact time b), c), d), the nuclide reaches farther and farther out into the water. When the water has passed the canister e), it does not take up any more nuclide.

65

Figure 1

Figure 2

rr : I , .

EI 0: 0, LOi

,/

/' ,/

/'

material

buffer E ial ~ v

rods embedded n lead ------

copper canister

Repository concept

, I

Repository as proposed in the Swedish Nuclear Fuel Safety study (KBS 1983). /20/

NINo \

0.1

0.05

004

0.03

0.02

0.01

--t--/

V /

/ I

2 3 4 5

Breakthrough curve for Pu239 . D a

66

\ \;

"-'\ \

buffer

1\

\ \ ~\ \\ time year

2 3 4 5

0.375 ffi.

Figure 5

Figure 6

Backfill .;".,'.!i

Canister f~~, [;'7~

Shows a crossection of the hole with a canister.

CLAY

H 20 2 COMPLETELY PARTIALLY PRODUCTION OXIDIZED OXIDIZED REDUCED

ZONE ZONE ZONE

Propagation of the radiolysis front downstream of a canister. The hydrogen peroxide and other oxidizing substances move with the water in the fractures, but also diffuse from the fractures into the rock, where the bivalent iron is oxidized to trivalent iron.

67

Figure 7

Figure 8

FLOW DIRECTION

• DISTANCE ACROSS THE FLOW DIRECTION

CONCENTRATION REDUCED

C SOLUBILITY

CONCENTRATION IN OXIDIZING ZONE ADIS,.ANCE DOWNSTREAM

PRECIPITATED QUANTITY

The concentration of a nuclide in the water in a fracture plane. A radionuclide with high solubility under oxidizing conditions pre­cipitates at the redox front. The water downstream of the redox front has a low concentration c~~1, but the front has broadened and a larger water flow transports the nuclide.

10-1

10-2

10-3

103

........ ......

23SU

"

\ 239pI.I

93Zr

'" )3scs I

99Tc ~ 237Np ,

1291

>-- 242PI.I

23SU

~8U

\-----........ ~

\ \ 1 OS 107 10S 109 1010

TIME AFTER CANISTER PENETRATION; YEARS

Release of radionuclides from the near field to the flowing water in the far field.

68

SOURCE-TERM CONSIDERATIONS FOR A POTENTIAL RADIOACTIVE-WASTE REPOSITORY LOCATED IN UNSATURATED TUFF

J. W. Braithwaite and M. S. Tierney NNWSI Repository Performance Assessments Division 6312

Sandia National Laboratories Albuquerque, New Mexico 87185

ABSTRACT

The Nevada Nuclear Waste Storage Investigations project is studying the feasibility of locating a repository for high level radioactive waste in the unsaturated tuffs at Yucca Mountain near the southwest part of the Nevada Test Site. An important part of these studies is the formulation of physically appropriate source terms for use in mathematical models used to assess the performance of the potential repository. Performance assessments conducted to date have used a preliminary source term based on simple assumptions and currently available data; this source term is described, along with the results of a sensitivity study that show the important parameters affecting the time-dependent radionuc1ide release rates.

liThe submitted manuscript has been authored by a contractor of the United States Government under contract. Accordingly, the United States Government retains a nonexclusive, royalty-free license to publish or reproduce the published form of this contribution, or allow others to do so, for United States Government purposes. II

69

INTRODUCTION

The Nevada Nuclear Waste Storage Investigations (NNWSI) project is studying the feasibility of locating a repository for high-level radioactive waste on or near the Nevada Test Site (NTS) in southern Nevada. Currently, the project is evaluating a site at Yucca Mountain near the southwest part of the NTS. The host rock in which the waste materials would be placed is a densely welded, fractured tuff located entirely in the unsaturated zone. Recent hydrological modeling studies (Sinnock et a1., 1984) suggest that the flux of water in the host rock is very low and is confined to the rock matrix. Water flux values ranging from 0.001 to 0.1 mm/Yr are expected and are consistent with laboratory measurements of unsaturated permeabi1ities (Peters et a1., 1984). The unsaturated, low-water flow condition is a distinguishing characteristic of this particular site.

Sandia National Laboratories (SNL) is assessing the postc10sure performance of the waste-isolation system at Yucca Mountain. In order to determine compliance with the various isolation and containment regulations that will be imposed on any geologic waste-disposal system by the U.S. Nuclear Regulatory Commission (NRC), the capability must exist to analyze the performance of the various system components at several different levels of detail. For example, an analysis of the time to failure of a waste container must be performed using space and time scales that are smaller than the ones used in the determination of groundwater travel time to the the accessible environment. To develop the needed capability, the performance assessment task at SNL is divided into two efforts: 1) an assessment of water flow and radionuc1ide transport through the site; and 2) an assessment of radionuc1ide release from the total system. The former i nvo1 ves two- and possi bly three-dimensional analyses of the hydrologic and transport characteristics of the near- and far-field geologic and geochemical settings. The latter involves simpler, largely one-dimensional studies of the total system; this less-detailed capability is required because of the expected large number of statistical sensitivity/uncertainty studies that must be performed to demonstrate system performance against long-term isolation standards.

In order to include radionuclide transport in either the site or the total-system assessments, descriptions of the rate of release of radionuclides as a function of time to the near-field geologic setting are required. These descriptions, or source terms, may be either phenomenological or empirical and will not, in general, be the same for both assessments. The radionuc1ide release from the waste package and the transport characteristics of the host rock surrounding the waste package must be determined to identify the quantity of rock (and thus the boundary) that must be accounted for in each source-term expression. For example, if the physical model in the total-system assessments includes only isothermal, single-phase flow, its source term must account for the transport processes occurring in the thermally and mechanically disturbed host rock surrounding the waste package. On the other hand, the multidimensional site assessments may include non-isothermal effects, and the source term in this case could then be the time-dependent release from the waste package.

A description of radionuclide release from the waste package is a crucial ingredient of any source term because, as mentioned in the preceding paragraph, this description must be included in the source terms used in

70

both the site and total-system assessments and is necessary for determining regulatory compliance with releases at the boundary of the engineered­barrier system. The Lawrence Livermore National Laboratory (LLNL) is designing the waste package and determining its detailed preclosure and postclosure performance, but it will be several years before the laboratory has collected sufficient data to allow these detailed analyses to be completed (Oversby, 1983; McCright et al., 1983). Because several current NNWSI activities require system performance assessments, (e.g., environmental assessments, data priority studies, site-characterization plans), an effort was undertaken at SNL to develop a source term that uses available data. A brief description of the analysis used in the development of this preliminary source term is included in the next section. A more detailed description of this procedure is contained in Braithwaite (1985). The third section gives the results of a sensitivity study that shows the effect the independent parameters included in this term have on the time-dependent release of radionuclides.

CURRENT NNWSI SOURCE TERM

APPROACH AND ASSUMPTIONS

In this section, the approach and assumptions employed in developing the preliminary source-term model currently used in the total-system-assessment codes are given. Principally, the development involved the identification of three items: the appropriate waste formes), the definition of the waste-form environment, and the appropriate description of the process controlling radionuclide release rates. A brief justification for each of these items is given separately below. It is very important to note that the source term described in this paper was developed for use in the system codes from information available in the spring of 1984. The level of detail is considered to be consistent with the detail of the system transport code (Sinnock et al., 1984) and, as such, does not represent the current state of knowledge. For a current status of waste package studies, see the paper by V. Oversby and D. McCright (1984) in these proceedings.

Waste form

According to current plans, the first commercial radioactive-waste repository will receive spent fuel; it may also receive a small quantity of either reprocessed defense waste or commercial waste (or both). The source term described in this paper considers only spent fuel.

Radionuclide transport and water availability

The hydrological information currently available provides estimates of the quantity and qualitative path of water flowing into and out of the repository horizon, but does not specify the water movement through the waste package or through the thermally and mechanically disturbed host rock surrounding the emplacement borehole. The characterization of the latter hydrological processes is the objective of several ongoing studies at SNL and LLNL.

Without detailed near-field hydrological information, there can be no analysis of the release-rate-reducing effect of the transport of radionuclides in solution from the waste form to the edge of the borehole

71

wall and then through the host rock to the appropriate boundary of the source term. Therefore, it was assumed in this source term that the transport of dissolved species in solution from the waste form to the borehole wall is fast and that the transport characteristics of the disturbed rock surrounding the waste package are equivalent to those for undisturbed rock.

To bound the possible saturation states of water within the waste package, both saturated (inundated) and partially saturated conditions were considered. Because the conductivity of the formation would be high under saturated conditions and because the natural state of the formation is unsaturated, conditions that would allow the w~ste package to become inundated would not be expected. However, before the waste canister fails in general, it is conceivable that a hole could develop in the top of it that would allow the inside of the canister to fill with water. Another reason for including the saturated environment was to make possible an assumption that the entire spent-fuel surface h~d beco~e wetted. Since dissolution rates can be dependent on the quantity of wetted surface area, this latter assumption allows any surface-area effect to be bounded.

The rate of water influx to the waste package was calculated by assuming that all downward-percolating water intersecting the horizontal cross­section of the emplacement borehole would enter the borehole. Results of stUdies of the effects of backfilling drifts on water-flow patterns support the conclusion that this calculational procedure leads to conservatively high rates of water influx (Fernandez and Freshley, 1984). Further, all water flowing into the borehole was assumed to contact bare spent fuel. Thus, the inhibiting influence of air gaps (up to 70% void volume can exist within the borehole), claddings, or other waste-package components on water contact with the spent fuel was not included (except, as will be noted below, to delay the onset of dissolution until lower temperatures exist).

Environment

The groundwater compOSition used in the solubility limit and applicable leach-rate calculations was taken from water produced by a well (denoted J-13) located on Jackass Flats near Yucca Mountain. This composition is given by Kerrisk (1984) and is currently used to represent the vadose water within Yucca Mountain. Additionally, the groundwater was assumed to maintain equilibration with air at a partial pressure of 0.2 atm of oxygen; this condition produces an oxidizing electrochemical potential in the pH = 7 groundwater of apprOXimately 700 mV.

The temperature of the spent fuel was always 25°C. This is a reasonable assumption because water should not contact the spent fuel until significant cooling has occurred and the activation energy for the spent-fuel leach rate is small (Braithwaite, 1985; Thomas, 1983; Johnson et al., 1981).

Radionuclide Release Mechanism

Congruent dissolution of the uranium-dioxide matrix is the mechanism used in this model to control the availability of radionuc1ides to the leaching groundwater. With the assumed uniform distribution of radionuc1ides throughout the spent fuel, this mechanism restricts their fractional rate of dissolution to that for the uranium-dioxide matrix.

72

Because of this restriction, this model is not capable of describing releases for species that are not contained within the matrix, such as portions of cesium and iodine, which can migrate to the spent-fuel gap, or carbon, which can exist outside the spent fuel. A few reasons why this approach was deemed reasonable are given by Braithwaite (1985).

IDENTIFICATION OF A DESCRIPTION OF THE RATE-CONTROLLING PROCESS

The dissolution or leach rate of the uranium-dioxide matrix is, in general, a function of time, the aqueous environment, and the chemical and physical properties of the spent fuel. Leach rates are typically determined experimentally in aqueous solution replenished often enough to ensure that any rate-retarding effects of soluble product buildup do not occur. However, if the leach rate is fast and little water is available, the concentration of the solubilized product can actually increase to the solubility limit. At this point, no further net dissolution will occur. Thus, saturation of the available water yields an upper bound on the dissolution rate.

The low-water-flux, unsaturated conditions expected at Yucca Mountain appear suitable to allow either solubility limits or leach rates to control the dissolution rate. That is, if water contacts the waste for relatively long times (e.g., in a water-filled container with breached areas only in the top), solubility limits coupled with water availability could control the rate. On the other hand, if the water-contact time is small (e.g., if a thin film or droplets flow over individual spent-fuel rods under partially saturated conditions), leach rates may control the dissolution. A separate analysis of potential water-flow patterns and associated contact times under both the expected partially saturated conditions and the unexpected short-term saturated conditions was performed. A description of this analysis is given by Braithwaite (1985). Part of this analysis showed that, for partially saturated conditions, water flow over vertically emplaced spent fuel should occur in discrete droplets with the surface-area coverage at any time being given by the following equation:

where SA F A

c

SA = C • F • A ,

= surface area covered per waste p-ackage (m 2/pkg), = flux of percolating water (m3/mZ/yr),

(1 )

= effective horizontal intercept area of groundwater by the package (m2/pkg).

= factor accounting for the residence time of the droplets and their ratio of contact area to volume ratio (2 x 10-4 yr • m2/m3).

The congruent leaching mechanism allows the analysis of the rate-controlling description process to focus on the dissolution rate of the uranium dioxide matrix. Because the half life of uranium is 4.5 x 109 years, radioactive decay will not significantly affect the quantity of uranium available for leaching over reasonable time periods. Therefore, the time before dissolution can commence (containment time) is not an important rate-determining effect. For convenience in this analysis, the rate of dissolution of the uranium matrix is defined as the mass fraction of uranium dissolved per year. This definition is related to the curie release-rate

73

expression used by the NRC when the total available mass is taken to be the inventory in the waste package 1000 years after closure. The mass­fraction-based release rate for the uranium matrix is given by equations (2) through (4) below.

For the dissolution rate controlled by the solubility limit and water availability:

where Rm

Sm f4m( 1 000)

Rm = F· A • Sm/Mro (1000) ,

= fractional release rate of the uranium-dioxide matrix based on the 1000-year inventory (l/year),

= solubility limit of the uranium-dioxide matrix (kg/m3) = mass of uranium remaining in the waste package at 1000

years after closure (kg).

Equation (2) is applicable to both saturated and partially saturated conditions.

For the dissolution rate controlled by the leach rate:

Rm = LR • SA/Mm (1000) ,

where LR = leach rate of the uranium matrix (kg/m2/yr).

(2)

(3)

Equation (3) is directly applicable to the constant wetted surface area or saturated conditions under which the leach-rate measurements are made; however, equation (1) can be substituted into equation (3) to give an expression defining release rate under the partially saturated conditions in which exposed surface area may not be constant:

Rm = C • LR • A • F /Mm (1000). ( 4)

To determine which of these two descriptions limits the rate of dissolution under the assumed environments, site-specific information is required concerning the leach rate and solubility limit of uranium dioxide. An initial uranium leach rate of 0.4 kg/m2/yr and a steady-state, long-term rate of 7 x 10-4 kg/m2/yr were selected as being the most representative results of those available in the literature for the expected environment (Braithwaite, 1985). Solubility limits for most of the radionuclides were calculated by Kerrisk (1984) using the geochemical code EQ3/6. For congruent leaching, only the total solubility limit for uranium is important (0.05 kg/m3).

Using equations (2), (3) and (4), one can determine the uranium dissolution rate for the two controlling processes as a function of the water flux. The results are plotted in Figure 1 for the saturated condition and in Figure 2 for the partially saturated condition. The two figures can be used to identify the process controlling the dissolution rate for both conditions. It is important to remember that an average water flux greater than 1 mm/yr is unlikely at the Yucca Mountain repository horizon; dissolution rates at a water flux up to 1000 mm/yr were included in this study to allow the dissolution rate to be shown under partially saturated conditions and steady-state kinetic control. The justification for the

74

selection of each of the rate-controlling descriptions is given in the following paragraphs.

Saturated conditions: As shown in Figure 1. dissolution based on the solubility limit is lower than, and therefore constrains the fractional rates until unrealistically high water fluxes are reached. At this point, the steady-state leach rate, which is not a function of the flux under these inundated conditions, begins to dominate. Therefore, if the canister interior or emplacement borehole becomes inundated,the water availability, coupled with the uranium solubility limit, should control the dissolution rate of spent fuel.

Partially saturated conditions: As shown in figure 2, leach kinetics should control the rate under partially saturated conditions. This is probably the case because of the short contact times between water droplets and spent fuel. While one expects the steady-state leach rates to dominate over the long term, the initial leach rate was initially selected as a conservative approach (by over two orders of magnitude) because the travel time for the droplets flowing over a spent-fuel rod (less than 10 seconds) probably represents the minimum possible value. Surface contamination, rod-to-rod contact, and the presence of discrete joints between individual pellets could increase droplet contact time and therefore the rate of dissolution.

If the actual leach rate were another 2.5 orders of magnitude higher than the assumed controlling initial leach rate, then solubility limits and water availability would constrain the dissolution process even under partially saturated conditions. With the uncertainty in the leach-rate data, water flow patterns within the waste package and water contact time, the choice of the solubility limit as a rate-controlling process is currently justified. Because the quantity of water flowing through the waste package does not depend on whether the local environment is fully or partially saturated, the dissolution rate based on solubility limits are identical for both saturated and unsaturated conditions.

SOURCE-TERM MODEL

The assumptions and analyses detailed above lead to a formulation of a preliminary source-term model. Three principal assumptions are inherent in this model: 1) the fractional rate of release of radionuclides from spent fuel is equivalent to the fractional rate of leaching of the uranium dioxide matrix; 2) the rate of dissolution of the uranium matrix is controlled by the solubility limit of the uranium coupled v/ith the availability of \'Iater; and 3) the transport of dissolved species to the borehole wall rock is instantaneous, and the transport characteristics of the thermally and mechanically disturbed rock around the waste package are similar to undisturbed rock. It must be remembered that the congruent leaching assumption (#l) is not applicable to the entire inventory of several radionuclides, including cesium, iodine, and carbon (Oversby and McCright, 1984) •

The preliminary source-term model can be described by Equation (5):

Ri(t) = F • A • Pr{Tc ~ t} • (minimum of Sm/Mm' S/M i ) (5)

75

where = The expected fractional rate (l/year) of release of the ith Ri (t)

species based, in general, on the mass inventory at the time of release (i.e., Mm or Mi could be time-depend~nt). For determining regulatory compliance, Mm or Mi is replaced with the mass inventory at 1000 years after closure.

F = Specific discharge of groundwater (~erco1ation flux) in vicinity of the waste package (m3/m /yr).

the

A = Effective area of intercept of groundwater by package (m2) •

Sm = Solubility limit of the ur~nium-dioxide matrix ~kg/m3). S' . = Solubility limit of the it waste element (kg/m ). 1 Tc = Time of container breach by water; i.e., the earliest time at

which water can freely enter and leave the interior of the container (years); a random variable.

Mm = Inventory (mass) of the urqnium~dioxide matrix contained in the waste package at the time of dissolution, or at 1000 years if the NRC criterion is being determined (kg).

M' = Inventory (mass) of the ith waste element contained in the 1 waste package at the time of dissolution, or at 1000 years if

Pr{Tc<t} the NRC criterion is being determined (kg).

= The cumulative probability distribution function of Tc• t = Time (yr) after closure of repository,

The minimum of the ratio of solubility limit to mass for either the matrix or the individual radionuc1ide is used in equation (5) to account for the precipitation of any species whose solubility limit would be exceeded under the assumption of congruent leaching.

SENSITIVITY OF RE~EASE RATE TO SOURCE-TERM PARAMETERS

For purposes of comparing the release rate to the NRC regulatory criterion, the mass inventory, M, must be set at its 1000-year value in equation 6. By ignoring mass loss by dissolution up to 1000 years and assuming that the ith waste element is a single-member decay chain, the following equation is obtained:

where, T 1 (2( i ) rvti 0)

= the half life of the ith species (years), = the mass inventory at closure of the ith speCies (kg).

The combination of equations (6) and (7) will be called the I'regulatory release rate. 1I

The cumulative probability distribution of container breach times is unknown; we assume it can be fit ~ith a two-parameter distributiqn with

(6)

arbitrary variance, 0'2, and mean Tc = air, where a is the container wall thickness (m), and Y' is the mean corrosion rate (m/yr). By defining the distribution of breach times in this way, the effects of spent-fuel cladding qn delaying or limiting the dissolution are explicitly ignored. Therefore, Tc is not a measure of the containment time specified in the NRC regulatory criterion.

76

The relative importance of the eight parameters appearing in the expressions for the release rate, equations (5) and (6), can be assessed by calculating and comparing the "normalized sensitivity coefficients" of the parameters relative to the release rate. The meaning of a normalized sensitivity coefficient is as follows: if ~ quantity R is a function of N parameters P1, P2, ••• ,Pn, the normalized sensitivity coefficient, Ci' of R with respect to the ith parameter Pi is defined by

Ci = (Pi /R}(aR/api ).

Evidently, Ci is a measure of the effect of a small, relative change in parameter Pi on relative changes induced in the dependent quantity, R. For example, a Ci = -1 means that a 10% increase in Pi would cause a 10% decrease in R if all other parameters remained constant.

(7)

The normalized sensitivity coefficients of the eight parameters appearing in equations (5) and (6) have been calculated or estimated and are listed in Table I. The CilS for all parameters except container-wall thickness, mean corrosion rate, and variance in container-breach time are exact and apply for arbitrary values of time and the eight parameters. The Ci IS for the three parameters, a, r, and a, that appear in the cumulative probability distribution of container-breach time are order-of-magnitude estimates and apply at times near the mean character breach time, Tc; these last three sensitivity coefficients were calculated using a piecewise-linear approximation to a general two-parameter distribution that was devised to simRlify the calculations. Because all cumulative probability distributions are similar, differing only in their shape parameters, it is likely that sensitivity coefficients derived from a physically motivated, continuous distribution (such as the inverse Gaussian distribution) would differ from the results shown in Table I only in the magnitude of the numerical constants (1/2 and 3 in the present case) that multiply the mean and variance; in most cases, these numerical constants are numbers between 1/2 and 10, and the qualitative features of the results in Table I would not change.

The sensitivity coefficients listed in Table I show that water flux, intercept area, solubility, and the square root of the variance (or standard deviation) in container-breach time are equally important in determining the release rate; the magnitudes of the sensitivity coefficients for these parameters are all 1, a value that can be taken as establishing a reference level for "importance." Thus, radionuc1ide half life is relatively important only for those radionuc1ide species with half-lives less than 693 years: in practical terms, variations in magnitude among the regulatory release rates for short-lived species. such as Cesium-137. are likely to be caused as much by differences among species half-lives as by differences among the species solubilities. On the other hand. variations in magnitude among the regulatory release rates for long-lived species such as llranium-235 are almost certainly caused by differences among the species solubilities if other parameters remain unchanged.

77

Table I Normalized Sensitivity Coefficients of Parameters in the

Regulatory Release Rate Expression

Pa rameter Name Water flux Intercept area Solubility limit

Initi a1 mass Standard deviation for T

Radionuc1ide half-life

Container-wall thickness

Mean corrosion rate

Sensitivity Coefficient

1

-1

-693/T1/2(i)

negative of wall thickness term

Effect of 10% Change on release rate

10%

-10%

-230% (cesium) -1 % (ameri c i urn )

. -0% (uranium)

- 9% ( T c I a = 3)

-87% (f cIa = 30)

negative of wall thickness term

Container wall thickness and the mean corrosion rate are equal in importance. However, these two quantities are importan! relative to water flux, intercept area, and solubility only if the ratio Tcla exceeds a number of the order of 10. If a is taken as a measure of uncertainty in the rate at which uniform corrosion attacks the material of the waste canister, exi sti ng knowl edge of that rate suggests that Tcl a is a number of the order of unity. For example, the lifetime against uniform corrosion of 1 cm of 304L stainless steel in the environment of the Yucca Mountain repository is estimated to 1 i e between 3000 and 30,000 years \'Ii th an expected val ue of about 10,000 years (McCright et al., 1983). Assuming_that the 27,000-year spread represents 4 standard deviations, the value of Tcla would equal 1.5, and the sensitivity coefficients for container-\'1all thickness and mean corrosion rate would be of the order of 1 or less. This example, which is based on the limited information available, suggests that container-wall thickness and mean corrosion rate may be slightly less important in determining the time-dependent release rate than water flux, intercept area, and solubilities. Reduction of the uncertainty in the types and rates of mechanisms that degrade waste-canister materials would be necessary before a different conclusion could be drawn.

78

SUMMARY

1. Source terms, or descriptions of the rate of radionuclide release to appropriate geologic settings, are required in total-system performance assessments, detailed site-performance assessments, and the determination of compliance with the criterion governing the release rate from the engineered barrier system.

2. A source term was formulated for use in the current total-system performance-assessment code that was based on information available as of the beginning of 1984. This term assumes that radionuclide release is controlled by congruent leaching and that the rate of uranium~diox;de dissolution is constrained by the uranium solubility limit coupled with the availability of water.

3. A study of the sensitivity of the regulatorY release rate to parameters in the preliminary source term was performed. Results of the study suggest that water flux, intercept area, waste-element solubility, and the standard deviation in container-breach time have equal normalized sensitivity coefficients. Sensitivity coefficients for the mean corrosion rate and container wall thickness are equal, but are likely to be smaller in magnitude than the coefficients for water flux, etc.; this conclusion is based on limited current information. .

REFERENCES

Braithwaite, J. W., "Effect of Water Flux on Uranium-Dioxide Dissolution in a Potential Radioactive-Waste Repository in Tuff," SAND84-1007, May 1985.

Fernandez, J. A., and Freshley, M. D., "Repository Sealing Concepts for the Nevada Nuclear Waste Storage Investigations Project," SAND83-1778, August 1984.

Kerrisk, J. F., "Solubility Limits on Radionuclide Dissolution at a Yucca Mountai n Reposi tory," LA-9995-MS, Ivtay 1984.

McCright, R. 0., et a1., "Selection of Candidate Canister Materials for High-Level Nuclear Waste Containment in a Tuff Repository," UCRL-89988, November 1983.

Oversby, V. M., "Performance Testing of Waste Forms in a Tuff Environment," UCRL-90045, November 1983.

Peters, R. R., K1avetter, A. E., Hall, 1., Blair, S. C., Heller, P. R., and Gee, G. W., "Fracture and Matrix Hydrologic Characteristics of Tuffaceous Materials from Yucca r~ountain, Nye County, Nevada," SAND84-1471, December 1984.

Sinnock, S., Lin, Y. T., and Brannen, J. P., "Preliminary Bounds on the Expected Postclosure Performance of the Yucca Mountai n Repository Si te, Southern Nevada," SAND84-1492, December 1984.

79

10-4

-:"'" 5 - 10-~ c .2 -u 10'" f! --a 10-7 'Q; ~ -

10-'

10-10

10-11

Iteady Itate kinetic

URANIUM Vertical emplacement Saturated conditions

1 0-12 '--.................... ..&.LI.&L..--'I.-.JI-L...&..&.Iu.u._-'--'-~u.LIu.---'.....L-'-u.a..a.u.._...I-""--I ..... .J.I.IJ

-... >-...... c 0 :;: u ca a.. --or; en 'Q; ~ -W .... ct a: z 0 i= :)

-' 0 f/) f/)

Q

10-2 10-1 100 101

WATER FLUX (mm/yr) Figu~e 1. The effe~t of water flux on the dissolution rate of

10-4

10-5

10-6

10-7

10-8

10-9

10-10

10-11

10-12

10-2

uranium dioxide under saturated conditions.

URANIUM Vertical emplacement Partially laturated conditions

100 101

steady state kinetiCs>.,.,.

."". WATER FLUX (mm/yr)

Figure 2. The effect of water flux on the dissolution rate of uranium dioxide under partially saturated conditions.

80

103

DATA REQUIREMENTS BASED ON PERFORMANCE ASSESSMENT ANALYSES OF CONCEPTUAL WASTE PACKAGES IN SALT REPOSITORIES

Gilbert E. Raines and John F. Kircher Office of Nuclear Waste Isolation

ABSTRACT

Performance assessment analyses of conceptual waste packages in salt reposi­tories were conducted as a part of statutory environmental assessments (EAs) for seven potential salt sites. These analyses included simulation of heat transfer, thermal gradient-induced brine migration, waste package corrosion, and dissolution of radionuclides from the waste and provided estimates of source terms for radionuclide migration based on available data. However, essentially none of the available data were site specific; these analyses identify explicitly those parameters for which site specific data are required to refine the estimates given in the EAs. These parameters are (1) thermal conductivity and diffusivity of host rock, (2) brine content of host rock, (3) composition and location of water bearing clay minerals in the host rock, (4) threshold thermal gradient for brine migration, (5) cor­rosion rates of overpack materials in site specific brines as a function of temperature and radiation level, (6) dissolution behavior of waste forms in site specific brines as a function of temperature and radiation level, (7) solute species and solubility levels as a function of temperature for various radionuclides in site specific brines, and (8) brine "diffusion" coefficients in salt. In addition, simulation models that are being used must be validated insofar as possible with tests in Exploratory Shaft Facilities. This paper will also report on the behavior of these brines as a function of temperature, radiation, and presence of package materials in the laboratory, using a version of the EQ3/6 geochemical equilibrium code which has been modified to accommodate high ionic strengths. The required precision for these parameters will also be discussed. The paper will provide details of these considerations.

(Abstract only available)

81-82

SUMMARY OF SESSION II: CONTROLLING MECHANISMS OF RELEASE

The papers presented in Session II and others of this workshop detailed a variety of mechanisms that may affect the release of radionuclides from nuclear waste packages. This summary divides these mechanisms into four broad categories of groundwater access, barrier penetration, waste form dis­solution,and radionuclide transport.

Groundwater Access

Several presentations (Braithwaite, Neretnieks, Pigford) demonstrated the central role of flow rate on the quantitative evaluation of radionuclide release rate in source-term calculations. Site-specific conditions for saturated flow (Neretnieks, Pigford), unsaturated flow (Braithwaite, Oversby), and brine migration/diffusive flow (Raines, Storck) were shown to lead to extremely different approaches to source--term modeling and testing. The role of groundwater access on source-tet~ evaluation is perhaps most clear for saturated host rocks such as granite, basalt, and clay. There is a continuous supply, albeit at low flow rates, of water to the waste package and a continuous water phase connecting the waste package to the surface. For salt host rock, the key aspect of groundwater access is the limited amount of water by corrosion of the metallic canister. In unsaturated tuff, the infiltration rate of groundwater is estimated to be extremely low and to occur as discrete droplets. There is nothing inherently superior in any of these flow conditions with respect to safe disposal of nuclear waste.

Another important feature affecting radionuclide release is ambient composition of ambient groundwater (Cleveland). Groundwater composition has been shown to affect both the rate of release and the solubility-limited concentration of released radionuclides. Complexation of radionuclides by dissolved organics should also be of concern to source--·term evaluations (Snelman). A caveat to this is the fact that ambient groundwaters can undergo significant changes in composition upon reaction at elevated tempera­ture with host rock and waste package materials (McKinley).

Barrier Penetration

Modeling the time-distributed failure of metallic barriers, rather than assuming simultaneous and total failure, may lead to significant attenuation of the release rate of radionuclides from a repository system (Wick, Rickertsen, Apted). Similar conclusions were reached by the KBS-3 analysis for 1291 release even for containment lifetimes of 10 6 years. It was also suggested that the limited rate of aperture growth, or exposure of waste form surface, may provide a limiting mechanism for radionuclide release (Apted), although the demonstration was qualitative and may prove to be difficult to explicitly evaluate.

83

Related to aperture growth, the suggestion was made that waste compo­nents and their alteration (corrosion) products will line the pathways of water entering a waste package. These materials will likely act as chemical buffers and reactive components to reduce radionuclide release (McKinley, Apted, Neretnieks). It was cautioned, however, that corrosion products often can effectively seal and protect underlying surfaces of fresh, unreac­tive materials such as metals. This would limit the enormous redox buffer­ing capacity represented by metallic barriers, but it should also retard overall corrosion rates.

Waste Form Dissolution

Source-term papers by a number of authors (Neretnieks, Pigford, Snelman, McKinley, Stelte, Wikjord) stressed solubility-limiting to release rate imposed by radionuclide--bearing solids. Three basic types of radio­nuclide release behavior were defined (Lanza, Avogadro): (1) nuclides that form their own solubility--limiting solids, (2) nuclides that are controlled by matrix-solubility limits of the matrix components of waste forms, and (3) nuclides that do not exhibit solubility-limited release behavior in any form (i.e., leach rate-limited).

Because of the heterogeneity of spent fuel, dissolution and release of radionuclides is unlikely to a~pear congruent because of the different dis­solution properties of the many phases (gap, grain boundaries, and grains) composing spent fuel. There is evidence, however, that U0 2 grains dis­solve congruently as a function of solution redox conditions (Shoesmith). Dissolution of borosilicate glass is congruent for time periods greater than several hours, although the subsequent precipitation of surface layer phases may indicate apparent incongruent releases (Wikjord).

Finally, it has been proposed (Neretnieks, Shoesmith) that alpha radiolysis at the waste form/water interface may lead to a locally oxidizing volume of solution. Because of the strong positive correlation between actinide solubility and oxidation potential (McVay), this highly oxidizing solution volume with relatively concentrated alpha-emitting nuclides may migrate away from the waste package. Interaction of such solutions with more reducing materials would eventually promote precipitation of nuclides. This scenario relies on complex interaction of surface area to volume ratios and relative rates of radiolysis, gas migration rates and capacity for redox buffering by solids, scavenging reactions within solution, and the reaction rate of solution redox couples.

Radionuclide Transport

The key role of groundwater flow rate on Qverall radionuclide release rate has been mentioned under "Groundwater Access." There are additional mechanisms that may further limit release. Reversible (physical) sorption within the waste package may attenuate the release of a limited number of nuclides that have either extremely high sorption coefficients, short half-lives, or both (McKinley, Pigford). Irreversible (chemisorption) sorp­tion, however, can be an effective mechanism for controlling the release of radionuclide in source-term calculations.

84

Formation and migration of radionuclide·-bearing colloids may also affect source--term calculations. Laboratory tests of waste forms (Lanza, Cleveland, McVay) indicate that a high proportion of many release nuclides, especially actinides, occur in a colloidal fraction. Initial work on the transport of such colloids through compacted clay indicates that essentially all the colloids are removed by physical filtration.

85-86

RELEASE MECHANISMS AND THEIR ROLE IN THE SPENT FUEL REPOSITORY PERFORMANCE

E. Peltonen*, M. Snellman, K. Ollila+ Technical Research Centre of Finland *Nuclear Engineering Laboratory +Reactor Laboratory Finland

ABSTRACT

The paper presents insights from studies almlng at clarifying the release mechanisms and their role in the overall safety of the final disposal of spent fuel in hard precambrian bedrock.

The expected chemical conditions are defined with respect to the materials introduced into the repository, taking into account their possible long-term changes. The main parameters affecting the barrier performance are identified. The parameters affecting the leaching and the solubility of the actinides in the repository are discussed basing on preliminary experimental and theoretical studies.

The role of release mechanisms for the safety of the overall disposal system under different conditions is discussed. The importance of the source term is increased when the conditions are assumed to be normal, e.g. the engineering barriers of the repository are damaged. The source term modelling requirements and the data needs are evaluated.

87

1. INTRODUCTION

The possible radiological impacts from final disposal of nuclear wastes can be estimated only by means of predictive mathematical models. An essential part of an overall model system is comprised of a near field modeling which describes the release of radionuclides out of the repository into the surrounding geological host formations. In that an important part consists of a "source term", which by definition describes the mobilization of radionu­clides from the wastes themselves.

In this paper the topic has been considered from the point of view of disposal concepts and spent fuel management timetables planned in Finland [1,2]. According to the present plans one is prepared to have a certain part of the spent fuel disposed of within Finnish territory after 2020. The R&D plan includes e.g. disposal site screening before the end of 1985, next safe­ty assessment for extension of the reactor operation licence by the end of 1987, preliminary site selection in 1992 and repository construction applica­tion in 2000. Alongside with spent fuel disposal R&D work an option based on the reprocessing and the disposal of immobilized reprocessing waste is under study, as well. This alternative, however, doesn't seem very probable due to unfavourable economics and lack of reprocessing capacity in the near future.

The present concepts concern disposal deep in geological forma­tions, in practice in crystalline rock such as granites, gneiss, gabbro etc., since salt or suitable clay formations do not exist.

The basic design of the repository consists of tunnels with verti­cal holes for the canisters containing the spent fuel. The disassembled spent fuel pins are enclosed inside copper canisters having a wall thickness of 0,1 ••• 0,2 m, a diameter of ca. 0,8 m, height ca. 4,5 m and weight 10 ••• 16 tons copper. The space around the pins is filled with lead or copper. The canis­ters are surrounded by compacted bentonite in the emplacement holes, the tun­nels and the shafts are filled with a mixture of quartz and bentonite. Detailed information about the repository and the encapsulation facility is presented in reference [1]. The repository and cani ster design may alter significantly, specially after the year 1992, when the site characteristics will be more restrictly bounded. Preliminary studies indicate sound reasons for a less sophisticated design [3,4].

In the R&D program the Technical Research Centre of Finland (VTT) plays an active role including experimental as well as modeling work, special­ly as regarding studies related to safety assessment.

2. MECHANISMS AND INTERACTIONS

The mobilization of radionuclides from the waste is effected by several interactions between the U0 2-matrix, water, engineered barriers and the surrounding rock: heat, radiation, chemical and mechanical effects. The mobilisation itself comprises dissolution of U0 2-matrix and its radionu­clides, diffusion of certain elements inside the matrix, and dissolution of gaseous elements from the gas gap of the fuel pins. The dissolution process includes interactive surface reactions and speciation and their solubilities in the water phase.

88

The mechanisms with their interactions should be taken into ac­count with a different accuracy at different repository development phases: one application purpose calls for very sophisticated modeling, another as simplified as possible, and furthermore, a realistic or a conservative ap­proach is to be chosen. So as to render all this possible, the mobilization phenomena and the pertinent phy~ico-chemical dependencies must be studied.

3. CHEMICAL CONDITIONS

In order to be able to assess the corrosi on of waste cani sters and the leaching and dissolution of spent fuel (U0 2 ) it is important to know the chemical conditions in the repository.

3.1 The bedrock/groundwater systems

Igneous rock, such as granite, is composed of a small number of major constituents e.g. quartz, feldspars (orthoclase, plagioclase), biotite, hornblende and pyroxenes. In addition there are some accessory minerals such as titanites, apatite, zircon, magnetite, hematite, pyrite, ilmenite, fluo­rite, chlorite, muscovite, epidote minerals and carbonates etc. that might have some influence on the composition of the groundwater.

The Finnish bedrock is more felsic (containing more quartz and feldspar) than the earth's crust in general. Smaller amounts of FeD, Fe Z03 , MgO and CaO and higher contents of Si0 2 and K20 are observed. Typical ot the Finnish bedrock is the high content of felsic granites and migmatites.

The primary rock forming minerals, additional accessory minerals as well as weathering and decomposition products form a very complicated system with the groundwater in bedrock. Some possible chemical effects of the main minerals in granites are listed in Table I.

In addition to these minerals also fluid inclusions might affect the rock-water equilibrium. In the STRIPA project [51 granite bedrock have been found containing fluid inclusions, up to some 10 8/cm 3 of rock. The fluid inclusions might contain very saline solutions, that is chlorides of Na, K and Ca. Some inclusions containing high concentrations of CO

2-gas have

also been found.

The composition of groundwater in granitic bedrock at greater depths is reported primarily in studies performed in Canada and Sweden. Most results from Finland are limited to the depths of shallow water « 200 m). Some trends in these data may, however,be extrapolated to greater depths.

The expected groundwater conditions in deep granitic bedrock as based on results available from Canada, Sweden and Finland are listed in Table II.

3.2 The repository/groundwater system

3.2.1 Conditions in the case of an intact canister

As long as the spent fuel canister remains intact, the conditions in the repository will be determined by the interaction of the groundwater

89

with the buffer and backfill materials. In addition the waste influences the chemical conditions mainly through temperature effects.

The effect of buffer and backfill materials

According to the proposed concepts the backfill consists of sand/bentonite mixtures, approximately 80 ••• 90 % of quartz sand and filler and 10 ••• 20 % of bentonite. The buffer bentonite Volclay MX-80 comprise 90 % of clay-minerals, 5 % of feldspars and 4 % of quartz. As accessories some calcite, biotite, chlorite, tremolite, titanite, apatite, talc, zircon, leucite, vulcanic glass, pyrite, ferro (ferri-) oxides and some silty sized rock fragments are to be met with. The content of organic materials and sulphide is expected to be low. The content of organic material in bentonite has been analyzed at the Technical Research Centre of Finland (VTT). The measured value was 5000 ••• 6000 ppm [6]. The sulphide content is according to KBS less than 200 ppm after oxidation [7].

The groundwater composition alters somewhat after passing through bentonite. The changes are mainly due to dissolution and ion exchange reac­tions between cations in the mineral and the groundwater components. At the early stages after closure bentonite contains some air, entrapped in the pores of the buffer mass, which will affect the 02-content of the water and redox conditions.

Due to temperature effects from the waste canister, the bentonite might undergo an illitization process, which in turn will influence the ion exchange reactions with groundwater.The effect of radiation is expected to be minor. Only a limited space between the canister surface and the bentonite is expected to be affected by the (y,~)-radiation.

Experimental

There are very few, if any, experimental results available treat­ing the effect bentonite has upon groundwater. Experimental studies on the bentonite-groundwater interactions have been carried out at the VTT. The groundwater used is a synthetic groundwater, the composition of which is given in Table III.

The groundwater represents a typical granitic groundwater well within the range given in Table II although some components such as Fe(II)/ Fe(III), sulphides, fluorides and phosphate are missing from the water.

The experiments are performed under static conditions at 250C in an anaerobic atmosphere (N 2 90 %/H2 10 %). Traces of oxygen « 20 ppm) are present in the anaerobic cabinet. Prior to starting the experTments the water is flushed with nitrogen (N 2). The level of oxygen in the water is measured at the start of the experiment and at the time of sampling. The oxygen level in the sample has remained low « 0,01 mg/l). The preliminary results of the experiments after 6 months are shown in Figure 1.

In the undisturbed deep groundwaters the H20-C0 2-CaC0 3 system is stated to buffer the pH to a range 6 ••• 10. In contact with bentonite, the original pH of the water (pH 8,2) rises to a higher level (pH 9 ••• 10).

90

The redox conditions in deep groundwaters are taken to be deter­mined by the presence of ferrous minerals, like pyrite FeS, siderite FeCO , and goethite FeOOH and the magnetite Fe 04/hematite Fe

20

3 equilibrium. Tfie

redox potential, Eh, based on the magnetite/hematite and a goethite/siderite equilibrium is calculated to be (-0,44 ••• 0 V) for pH 6 ••• 10. The preliminary redox measurements performed with the Ag/AgCl and the glassy carbon elec­trodes in the bentonite-water system indicate a negative Eh-potential. However, the Eh-values for the samples range from -0,1 to -0,3 V. The reason for this wide range is not yet known. In addition to the Fe II/Fe III-couple several other redox couples, such as S(VI)/S(II), N(V)/N(I), 0(0)/0(-11) and organic components, might be present.

The principal anions in the groundwaters are OH, HC03

, CO , S04' Cl, F with minor amounts of HPO , P0 4, N0 3 , and HS. These anions migfit form complexes with the elements of the nuclear waste (canister and spent fuel). The carbonate concentration in the groundwater is strongly affected by the presence of bentonite, due to dissolution of carbonate minerals. Some disso­lution of fluoride and chloride from the bentonite is also observed. No measurable amounts of sulphide (> 0,1 mg/l) have been observed.

In addition to the inorganic ligands the concentration of com­plexing organic compounds (fulvic and humic acids) in the groundwater is of importance. The organic content of groundwater is generally 1 ••• 20 mg/l, the fraction of fulvic and humic acids is not known. The increase in the KMn0 4 consumption in the bentonite-water experiments indicates some dissolution of organic material from bentonite, or some bacterial (anaerobic) growth. The composition of the organic component has not been analyzed as yet.

The concentration of cations Ca, Na, Mg and K in the water is influenced by the bentonite due to ion exchange and dissolution effects.

A summary of the expected chemical conditions in the repository prior to canister breakdown has been given in Table II.

3.2.2 Conditions after canister penetration

After the canister has been penetrated the chemical conditions in the repository will in addition to the material effects mentioned earlier be modified by a-radiolysis and the dissolution of spent fuel and cladding materials.

The main radiolysis products are H and H20. The yield of these products depending on the amount of Fe(II), Hcd 3 , C0 3 ' C1, N0 3 and N0 2 in the groundwater. The net effect of these over time spans seems to be complicated to calculate.

Christensen and Bjergbakke calculated the radiolysis effects for water with a low iron content, < 5 ppm. No further addition of iron e.g. from bentonite was considered. The production after 10 6 year totaled 41 550 mol for 1,4 tons of spent fuel. Irradiation was considered in a 4,2 1 water volume in a layer 0,03 mm thick [9].

91

Due to radio1ysis effects the metals Zr, Pb and Cu are expected to be oxidized first by leaving the actinides in their original oxidation states. The metals form hardly dissolvable salts and oxides, such as PhO, PbS0 4 and Zr02 • The formation of protective layers of oxides will most like­ly decrease tne effectiveness of the metallic reducing agents, leading to an increased oxidation of the spent fuel and higher mobilities of the actinids.

In addition to the Eh-effects the radiation will also influence the pH conditions in the repository. The extent of the effect is not known.

Table II lists the expect'ed groundwater conditions at the time of dissolution of the spent fuel (U0

2).

4. FACTORS AFFECTING THE DISSOLUTION AND LEACHING MECHANISMS

Spent fuel is largely U0 2, with only a small fraction being con­stituted by other actinides and fisslon products. The chemical and physical changes that occur during irradiation result in a product that is inhomoge­neous on both the micro and macro scale. This inhomogeneity is reflected in the dissolution behaviour e.g. the dissolution of gas gap elements and ele­ments from the fracture surfaces.

4.1 Dissolution mechanisms

The dissolution of U02

as modified by the fue1's irradiation hi story can be descri bed as:

1) An almost instantaneous release of soluble nuclides of Cs, I present in easily accessible regions of the fuel pins either at the gas gap or in cracks or porous regions of the pellets.

2) A congruent dissolution of U0 2 , which releases other radionuc1ides, such as the actinides in the solid. This is complicated by a per­turbance due to the diffusion of certain elements inside the matrix.

Models for the congruent dissolution of fuel are usually based on the solubility of U0 2 in a limited amount of water. This solubility can vary by several orders of magni tude dependi ng on the redox chemi stry and ground­water composition. Under oxidizing conditions U0 2 solubility is stated to be limited to about 360 mg/1 by the availability of carbonate (about 275 mg/1) [10]. This conclusion is based on theoretical calculations.

The radi olysi s may affect the behavi our of the radi onuc1 i des vi a effects on the groundwater, that is mainly Eh and pH changes, or by the radio­lytic conversion of crystalline oxides to amorphous or less crystalline oxides thereby decreasing their thermodynamic stability and increasing their solu-bi 1 i ty.

Among the inorganic ligands generally found in groundwater HC0 3 , C0

3 and OH are of particular interest owing to their high concentrations as

well as to the high stability constants of the complexes they form with actinides.

92

Experimental

The mechanism of spent fuel U0 2 dissolution has generally been studied using unirradiated U0 2 pellets. So far only some experiments have been performed with spent fuel. Most of the studies have been carried out using deionized water. The mechanism will, however, be more complicated in groundwater conditions. Some suggested oxidation and dissolution mechanisms for U0 2 is shown in Figure 2 [11]

Electrochemical studies performed by Johnson et al. have elucidat­ed some aspects of the initial oxidative dissolution of U0 2 [121. The U0 2 surface reacts rapidly with dissolved oxygen to produce oXldlzed surface films. The thickness and composition of said films depend on the redox condi­tions and solution composition. The effect of oxidized surface films on the leaching and dissolution behaviour of irradiated fuel appears to be of signif­icance, but adequate information on the subject is not to be had at present.

Irradiated fuel dissolution experiments at 2S oC have shown that under oxidizing conditions the U0 2 matrix dissolution rates can be enhanced by the presence of complexing agents, such as carbonate. In spite of this result, fission product and actinide releases have been reported to be gener­ally lower in groundwater than in pure water. The reasons for this are not clear but could be explained by processes occurring after the release of the U0

2 species from the fuel i.e. adsorption onto the fuel surface or colloid

formation and precipitation processes [12J.

Experimental and theoretical studies on the dissolution and leaching behaviour of spent fuel under repository conditions has been studied at the VTT since 1980. Unirradiated U0 2 fuel is used in the experiments. The tests have up till now been performed in natural groundwaters taken at shallow depths (< 70 m) in the Finnish bedrock and at the Stripa mine in Sweden. Recently a series of dissolution experiments with artificial groundwaters was started, the one representing a typical granitic groundwater (Allard) and the other corresponding to the repository conditions taking simultaneously into account the possible effect of bentonite (Table II). In these experiments the pH and redox conditions are varied.

Up till now the dissolution tests of unirradiated U02

pellets have been made using the ISO (International Standard Organization) standard method to measure the dissolution rate of uranium in groundwater. The method was modified so that the low oxygen content of groundwater was simulated. The experiment was carried out in an oxygen-free atmosphere in an anaerobic cabinet filled with a mixture of nitrogen and ~drogen (90 % N2/10 % H2 ). Two different natural groundwaters were used as leachants. The composltions are presented in Table IV. The Olkiluoto groundwater has a relatively high content of HCO , which is near the upper limit specified for groundwaters in deep granitic 5edrock (Table II). The concentration of the main groundwater components is generally smaller for the Stripa groundwater, except for chloride and fluoride.

93

The groundwaters were made oxygen ... free by fl ushi ng with N 2 gas. Then the samples with U02 pellets were removed into the anaerobic cablnet. The oxygen content in the waters remained at ~ 0,01 ppm during the experi­ments. The accurate content could not be measured. The redox potential values of the waters ranged from -0,4 ••• -0,1 V.

The dissolution rate of uranium in the natural groundwaters as a function of time up to 400 d is presented in Fig.3. The experiment is still continuing. The dissolution rate is 3 ••• 4-10- 8 g cm- 2 d- 1 after 520 d.

Duri ng be observed in the during the period. dissolution rates,

the last dissolution periods precipitation of uranium could waters. It seems as if the solubility limit were attained

When the solubility limit is calculated from the measured the following solubility limits are obtained:

Olkiluoto groundwater: 25 ••• 200 ~g/l Stripa groundwater: 40 ••• 350 ~g/l.

Moreover, an autoclave system for dissolution experiments has been planned and constructed, see Fig. 4.

The groundwater is warmed, and made oxygen-free in the supply vessel by flushing with N2 gas. Then it is let run down in the dissolution vessel where the experiment is carried out. The redox, pressure and tempera­ture conditions may be adjusted. The system has been constructed in a way that the U0 2 pellet does not undergo any variation in pressure during water changing. ~inally, the groundwater is let run down in the depletion vessel for sampling. Recently, a redox measurement system with a Pt electrode and a Ag/AgCl reference electrode has been installed in the dissolution vessel.

The first dissolution experiment with the autoclave system has been performed. The purpose was to test the system. The dissolution rate of uranium in natural groundwater (Olkiluoto groundwater) was measured at ele­vated pressure, 10 MPa, at 60 oC.

The dissolution rate as a function of time up to 300 d is pre­sented in Fig. 5. The value is 5x10- 8 g cm- 2 d- 1 after 300 d. The experi­ment was finished after that.

The prel imi nary experiments show that the temperature and the pressure do not significantly affect the dissolution rate of U0 2 fuel pel­lets. No significant difference between the effect of the two natural groundwaters (HC0 3 , S04' Cl, F) upon the dissolution rate is observed.

94

TABLE I. THE CHEMICAL EFFECTS OF GRANITIC MINERALS

Mineral /weathering product

Main minerals Quartz 25 - 30 %

Feldspar 50 - 75 % E.g. orthoclase KA1Si 30S E.g. plagioclase (Na,Ca)(Al,Si)40a

Amphiboles < 20 % E. g. hornblende Ca 2Na(Mg,Fe)4 (Al,Fe,Ti)3 Si S022 (OH,F)2 ,

Micas < 20 % !

E.g. biotite

K(Fe,Mg)3 A1Si 3010 (OH,F)2

Accessory minerals Magnetite Fe304 Pyrite FeS 2

Ilmenite FeTi03 Zircon ZrSi04 Fl uori te CaF 2

Calcite CaC0 3 Apatite Cas(F,Cl,OH)

(P0 4 )3

Monazite (Ce,La,Th)P04

Weathering products Kaolinite A1 4(Si 4010 )(OH) Chlorite (Mg,Fe,Al)4(Si 401o )(OH)a

Clay minerals

Chemical effect

Source of colloid 5i0 2

Interstitial K, Na, Ca-exchangable

Easily weathered

Cation exchangers?

Eh-effect, Fe 2+-source Eh-effect, Fe 2+-, 52-, 50 42-source

Partially exchangable (M(IV))? F--source Ca 2+-, HC0 3--, C0 32--source, pH Hp0 42--, P043-- source

Sorption of Ln and An?

From feldspars, cation exchanger

From amphiboles and micas,

cation exchanger

Na+, Ca2+, Mg2+, pH

95

5. DISCUSSION AND CONCLUSIONS

The importance of understanding the source term mechanisms depends on several factors, the most relevant possibly being:

the criteria to be applied in assessing the suitability of a disposal system

- the repository development phase.

When the system performance criteria are used, the role of the source term depends on the performance of the other parts of the whol e di s­posal system. If the geosphere is evaluated as reliable and effective, there is no need of high accuracy in the source term estimation. On the other hand, if the geosphere is either ineffective at present or if the probability of such disruptive events, which can weaken essentially its performance in the future, is high enough, the role of the source term'becomes emphasized. Particularly in situations, when the other engineered barriers of the reposi­tory may be breached, consequently increasing the water flow around the waste, the source term may become very important.

If the criteria concern the performance of subsystems or compo­nents, the role of the source term is evidently greater and more independent of the performance of the rest of the system. The level of understanding called for depends, however, on the strictness of the criteria.

If the repos i tory development is in the phase of concept feas i­bility assessment, the geosphere may be assumed to be rather ineffective, and thus the release mechanisms should be known well enough so that the conserva­tism of the source term estimate can be relied upon. After the site selec­tion phase, in a phase of engineered barrier design, the source term-related mechanisms must be known even better so as to be able to develop and apply realistic modeling for optimization purposes.

The uncertai nty estimates of the performance are of importance and they are to be made. The estimation methods and their contents are depen­dent on the type of criteria employed and of the repository development phase. An understanding of the mechanisms affecting the source term improves the reliability of the uncertainty estimates, regardless of whether they are made by means of sophisticated or simplified models.

The following research objects are therefore considered important in the near future:

Groundwater conditions inside the repository, in the surrounding rock and in the near field (disturbed zone), especially changes due to buffer/water chemical interactions and waste/water radiolytic interac­tions in a realistic repository environment. Dissolution mechanisms, speciation and solubility under realistic reposi­tory conditions. Identification of the most important dependencies on the various possible factors and quantification of the effects.

The phenomena and the connected mechanisms are to be studied by experimental and theoretical means. In the source term modelling the interactions will be considered more carefully and will be given more attention in the development work to be done.

96

[1 ]

[2 ]

[3 ]

[4 ]

[5 ]

[6 ]

[7 ]

[8 ]

[9 ]

[10 ]

[11 J

[12 ]

REFERENCES

Final disposal of spent nuclear fuel into the Finnnish bedrock. Report YJT-82-46, Nuclear Waste Commission of Finnish Power Companies. Helsinki, 1982.

Decision in principle of 10th Nov. 1983 by the Council of State of Finland on the objectives to be observed in carrying out research, surveys and planning in the field of Nuclear Waste Management. Helsinki, 1984.

Vuori, S., et al.: "Envi ronmental impact and ri sk analysis of direct disposal of spent fuel as compared to reprocessing", ANS/ENS Topical Meeting on Fuel Reprocessing and Waste Management, Jackson, Wyoming, USA, 26-29 August, 1984 (to be published).

Vuori, S., et al.: "Applicability of Cost-Effectiveness Approach for Comparison of Waste Management Options", Proc. SymtOSiUm on the Risks and Benefits of Energy Systems. Internationa Atomic Energy Agency, JUlich, FRG, 9-13 April, 1984 (to be published).

Lindblom, S., "Fluid inclusion studies of the Stripa granite", Stockholm, Sweden, February, 1984 (to be published in a STRIPA project report).

Holopainen, P., et al.: "Crushed Aggregate-Bentonite Mixtures as Backfill Materials for the Finnish Repositories of Low- and Intermediate-Level Radioactive Wastes". Report YJT-84-07, Nuclear Waste Commission of Finnish Power Companies. Helsinki, 1984.

Corrosion resistance of a copper canister for spent nuclear fuel. The Swedish Corrosion Research Institute and its reference group. KBS Technical Report 83-24. Stockholm, 1983.

Allard, B., et al.: IIMinerals and precipitates in fractures and their effects on the retention of radionuclides in crystalline rocks ll

, Proc. Near Field Phenomena in Geologic Repositories for Radioactive Waste. pp. 93-101, Seattle, 1981.

Christensen, H., Bjergbakke, E.: IIRadiolysis of groundwater from spent fuel". KBS Technical Report 82-18,1982.

Final Storage of Spent Nuclear Fuel - KBS-3, III Barriers. Swedish Nuclear Fuel Supply CO./Division KBS, Stockholm, 1983.

Wang, R., et ale Proc. MRC Sym~OSiUm Scientific Basis for Radio­active Waste Management. pp. 79-386, Boston, 1981

Johnson, L.H., et al.: IIMechanisms of Leaching and Dissolution of U0 2 Fuel II , Nuclear Technology, ~, 238-253 (1982).

97

TABLE II. CHEMICAL CONDITIONS IN THE REPOSITORY

Component Natural Repos i tory groundwater groundwater In case of After canister

intact canister penetration

pH 6 .•• 10 6 ... 10(11) 4(6) ... 10 Eh(V) -0,44 ... 0 -0,4 ... 0 -O,4 ••• pos.values Conductivity mS/m 30 ••• 500 O2 mg/l < 1 > 1 > 1

HC0 3 - mg/l 100 ••• 400 400 ••• 600 400 ••• 600

COl 2- .. S04 2- .. 1 ... 100 1 ••• 50 i 50 HS- .. 0,1 ... 1 1 .•. 3 < 1

HP0 4 2- .. < 0,4 ~ 0,4 < 0,4 ... P0 4 3- ..

~ 0,01 ~ 0,01 < 0,01 ... N0 3 -

.. 0,01 ... 0,05 0,01. •. 0,05 0,01. .. 0,05

NO 2- .. 0,01 ... 0,6 0,01 ••• 0,6 0,01. .. 0,6 C 1- .. 5 ••• 100 5 ... 100 5 ... 100 ,.. F- .. 0,5 ... 7,5 < 7,5 < 7,5

"" ... Ca 2+ .. 10 ••• 100 < 20 < 20 Mg2+ .. 1. .. 65 '( 5 '( 5 ... ... Fe tot

.. 0,02 ... 5 1 ••• 10 1. .. 10 Mn 2+ .. 0,02 ... 0,7 0,02 ... 0,2 < ... 0,2 K+ .. 0,2 ... 10 < 5 < 5

"" ... Na+ .. 10 •• 500 300 ••• 500 300 ••• 500

NH4 + .. < 0,1 < 0,1 < 0,1 ... ... AP+ II

~ 0,02 ~ 0,02

Cu, Zr, Pb .. compounds

Si0 2 .. 2 ••• 20 > 2 ••• 20 2 ••• 20

Org. compounds 1 ••. 20 30 ... 40 < 30 ... Colloids II 0,5 ... 1 > 1 > 1

"" ... Micro-org. .. Gases:

H 2 ' H20 2 mol/l < 360 < 41 500 "" ""

Temperature Oc 8 ... 13 max. 70 0 C 8 ... 13

98

TABLE IlL COMPOSITION OF THE SYNTHETIC GROUND­WATER ACCORDING TO B. ALLARD [8].

HC0 3 Si0 2 SO 4 Cl Ca Mg K Na

pH ~ 8,2

TABLE IV.

pH Conductivity (25 0 C) HC0 3 F NO 3 N0

2 NH 4 Cl SO 4 PO 4 Fe Mn Ca Mg Na K Si

Cone [M] mg/l

2 x 10- 3 123 2 x 10- 1+ 12 1 x 10- 4 9,6 2 x 10- 3 70

4,5 x 10- 1+ 18 1 ,8 X 10- 4 4,3

1 x 10- 4 3,9 2,8 x 10- 3 65

COMPOSITION OF GROUNDWATERS USED IN U0 2 DISSOLUTION EXPERIMENTS.

Olkiluoto ground- Stripa ground-water water

8.2 8.3 8.7 mS/m 28.0 mS/m 360 mg/l 86 mg/l 0.47 II 5.25 II

< 0.1 II 0.02 II

< 0.1 II < 0.001 11 < 0.0111 < 0.01 20 II 37 180 II 6.0 0.011 II < 0.1 0.04 II < 0.02 1.29 II < 0.02 69 II 13.6 39 II 0.25 77 II 50 17 II 0.26 13 II < 1 (Si0

2)

99

..... .... HCb mg/l TVO. BENTONITE AND SIMULATED GROUNDWATER

.... (ALLARD) ....... ......

CO 3 " ;; .... C1 3 " .... ....

.... Na "

........ .... .... S04 " .... ....

.... K~'ln04-CONS g/l ".-

/ I

CONDo mS/m I I

pH /

/ HC03 I

I 300 /

/ I

I I

10 I /

Na I

/

200 . / pH -

5 100 COND!!CTIVUY • •

~ __ -~ _____________ ~l _________ . __ .. . -=---=-... __ 4- - _ KMn04 ________ "_ ...

::",.>c::.. ---- C03" -4-- -- - _ S 4 -- 11- _____________ -)(

30d gOd l80d Figure 1. Changes in groundwater composition due to bentonite.

Figure 2.

UO HX

AQUEOUS PHASE

UOJ

IUO;J SOURCE)

... ~ .•. ; ~.O:·:·.~. r.

o 0" ,;,<)~. ,.0, '. .' . () " O <)'0' .'

•. .' -'.0 ~---E----~ UO:' ----~.~ • •

(\1\: 00· V V· '" ,", • o' , ,. II • . .. '" " .. '0 . 0 . n"

t o;'2 ~t,.;o .,) ". o· .... ".

t PARTICULATES. POLYMERY

'" /PLATING COATI~ UO, HYDRATES

CONTAINER WALL OR SOLID SURFACE

+I IUOJ SINK)

Oxidation and dissolution mechanisms for U02 [11].

100

LEACHANT

~ ___ FLUSHING WITH N2

SUPPLY VESSEL

- PRESSURE

1--1-_~\7L. BALANCING

- DEOXYGENATION

DISSOLUTION VESSEL PRESSURIZATION ----.01----1 WIT H N2 1---1--"'--" ..

Figure 4.

• 11 - _LEACHANT

U02 pellet

DEPLETION VESSEL

-PRESSURE BALANCING

SAMPLING

Dissolution at elevated pressure.

101

o 100 200 JOO 400 Natural groundwaters ~~-r-r~~~~'-~-T~--~~~-r~~~~~~~ 25°C

o N 1: 1 SO 1-+-----------1,.-------__+-----1 o '-o (Xl

o

I:9-t!l 2 8 - 1 ~ 28-2 ~ 2C-l +-+ 2C-2

1---l150

~ 100 H-------If------__+-------l-------I 100

W I­< 0:: 50 ....... +------I--+-~-__+--------_+------:....j SO J: o < W ....J

100 200

TIME 0

o JOO 400

Figure 3. U-disso1ution rate as a function of time (25°C).

05 N 1: o ...... (!)4

~ o I WJ

W 1-2 < 0::

:I: 0 1 < W ....J

o

o

1

o

100

D

\

\'" m~ 100

200

- ..

- --200

TIME 0

JOO

5

4

J

2

..

o JOO

low oxygen oontent Eh (-0.4 ••• -0.1)v

Natural groun:'iwater 60°C low oxygen oontent 10 MPa

Figure 5. U-disso1ution rate as a function of time (60°C, 10 MPa).

102

LEACHING OF PLUTONIUM FROM A BOROSILICATE GLASS BY SELECTED GROUND I-JATERS

T. F. Rees, J. M. Cleveland, and K. L. Nash U.S. Geological Survey Denver, Colorado U.S.A.

ABST'R.ACT

The leachability and speciation of plutonium from a borosilicate ~lass (Battelle 80-207) by eight actual ground waters of diverse chemical composi­tions has been determined. Waters from alluvial deposits in the Hualapai Valley (Arizona) and from Grande Ronde Basalt (Washington) leached plutonium most efficiently at 2SoC and 900 C respectively. Leaching was incongruent with plutonium removed slower than glass-matrix dissolution. The results indicate the need to determine leachability of actual waste forms with actual ground waters that would contact the waste in the event of a repository breach.

103

INTRODUCTION

The U.S. Geological Survey program to study the geochemical behavior of the transuranium elements in ground waters from rock types that are being con­sidered as possible hosts for nuclear-waste repositories has involved three approaches: 1) single-phase speciation of added ulutonium, neptunium, and americium in 12 ground waters and 2 surface waters as a function of time and temperature [1-7]; 2) determination of thermodynamic parameters for com­plexes with ions common in ground waters [8-10]; 3) determination of the leachability and speciation of plutonium from a radioactive waste glass by selected ground waters used in the single-phase speciation studies. This paper describes results of the latter study.

A theory for glass leaching has been formulated from existing data using deionized water as leachant. In this theory, the initial stages of leaching follow a diffusion process in which the alkali-metal ions are selectively leached from the surface of the glass, leaving behind a porous, silica-en­riched hydrated layer. As this dealkalized layer thickens, the metal dif­fusion rate becomes progressively slower until the diffusion rate equals the rate of dissolution of the silica-enriched layer. At this time, dissolution of the glass appears to become congruent. Barkatt and associates [11] ob­served that when polyvalent cations are present in the glass matrix, they tend to be retained in the dealkalized layer until all available sites in the layer have been saturated, after which they are released at rates much great­er than those measured initially. Thus the release rate of polyvalent ions should increase with increasing leach time.

One consequence of the theory is that the thickness and structure of the surface layer is probably dependent on the chemical composition of the leach­ing solvent. In fact, there were observable differences in the depth and structure of the surface layer in glasses leached by deionized water compared to glasses leached by brines [12]. However, few leaching studies have utili­zed actual ground waters, the leachants that would actually contact the wastes should a repository be breached [I, 13-15]. In this paper, the results of a study using eight ground waters to leach a plutonium-containing borosili­cate glass waste form are presented. We will show that the quantity, chemi­cal form, and physical form of the plutonium leached from the glass are highly dependent on the chemical composition of the ground-water.

EXPERIMENTAL

The waters used were from: 1) Grande Ronde Basalt underlying the Han­ford reservation in Washin~ton; 2) tuff at the Nevada Test Site (NTS); 3) Cretaceous shale of the Northern Great Plains; 4) Climax stock granite at the NTS; 5) Crystal Pool, ~"hich drains the carbonate aquifer underlying NTS; 6) well 5C at NTS, which nenetrates alluvial fill; 7) Rentfrow well, which penetrates into the Permian Yeso formation in the Tularosa Basin in New Mexico; and 8) Red Lake South well, \"hich penetrates alluvial fill in the Hualapai Valley in Arizona. Deionized water was included for comparison to other studies. Ground-water conryositions are shown in Table I.

The leaching studies uere conducted with 5-mm glass cubes (Battelle 80-207 [16]) and contained O. ')I:. Height percent plutonium. The chemical composi-

104

tion of the glass is given in Table II. All leaching experiments were conduc­ted using the Battelle static leach test MCC-l procedure [17] with two modifi­cations: 1) Only two temperatures were used (2S o and 900C); and 2) the sam­pling periods were 30, 90, 180, and 360 days. The surface-area-to~volume ratio (SA/V) requirement was met by using three cubes (area of lSOmm2 each) and 4S mL ground water contained in tightly sealed Savillex (1) Teflon jars. Weight loss from these jars averaged less than 1 percent, even after 360 days at 90

0C. After the stipulated leaching period, the waters were sampled before

and after filtration through O.OS-wm Nuclepore filters and the oxidation-state distribution of plutonium in the filtrates was determined by procedures given in [18], involving PrF3 carrier precipitation to determine Pu(III) and Pu(IV) con~entrations, P:F3 precipitation ~ollowing NaHS03 reduction f~r total plu­tonlum concentratl0n, and thenoyltrlfluoroacetone ~TTA) extractl0n for Pu(IV) concentration. Plutonium (V+VI) concentrations were calculated by subtract­ing the Pu(III) and (IV) concentrations from the total plutonium concentra­tion. In addition, the glass cubes and containers were rinsed with 0.5 ~ HCl04 , and the resulting solutions were analyzed to determine the quantity of sorbed plutonium. Because the plutonium was not isotopically pure and the glass contained a small amount of americium, all plutonium analyses required ion-exchange purification followed by alpha-spectrometric analysis as des­cribed in [19]. The concentrations reported are for Pu-239.

To determine whether the mechanism of leaching was consistent with that formulated by Barkatt [11] for deionized water, experiments at 25 0 and 90

0C

were performed utilizing plutonium-free glass cubes otherwise similar to those used in the plutonium leaching experiments. The chemical compositions of the three ground waters {basalt, shale, and tuff) used as leachants in these ex­periments were determined on aliquots that had contacted the glass for 180 days, and on aliquots receiving the same treatment in the absence of glass cubes. Changes in chemical composition between the two sets of waters were attributed to material leached from the glass.

RESULTS AND DISCUSSION

Results of the plutonium speciation experiments are shown in Figures 1 and 2. The data at each time period were obtained from a single experiment. The insoluble-sorbed fraction is the quantity of plutonium removed from the glass cube surfaces and the Teflon vessel walls by rinsing with dilute llCl0

4,

and represents the fraction of plutonium that was leached from the glass, but subsequently sorbed. The insoluble-suspended fraction is the difference between the total plutonium concentration in the ground water before and after filtration through a O.OS-wm filter, whereas the soluble fraction is the con­centration of plutonium that passed through the filter, and is defined as be­ing in solution, although colloidal material smaller than 0.05 wm might be present in this fraction. Within experimental error, which we estimate based on counting statistics, agreement of the individual oxidation+state analyses and previously reported [1-7] single-phase experiments to be - 20%, the sum of the concentrations in each of the oxidation states should equal the con­centration of plutonium in the soluble fraction.

The quantity of plutonium leached from the glass (soluble, insoluble­sorbed, and insoluble-suspended), its physical form and its oxidation-state

(1) The use of trade names is for descriptive purposes only and does not con­stitute endorsement by the U.S. Geological Survey.

105

distribution all depended on which ground water was used as leachant. At o

25 C, the basalt, Crystal Pool, Rentfrow and Red Lake South ground waters were about equally effective leachants, whereas well 5C maintained the most plutonium in solution. At 90oC, the basalt removed the most plutonium and maintained the highest concentration in solution. With the exception of the basalt and tuff ground waters, more plutonium was leached at room temperature. In general, more remained in solution after 180 days than after 360 days, in­dicating some mechanism for removing the plutonium fron solution once it was released from the glass. An auxiliary experiment in which both radioactive and non-radioactive glass cubes were leached together in the same Teflon ves­sel showed that the amount of plutonium rinsed off the non-radioactive cube was about 70 percent of that rinsed from the radioactive one, indicating that most of this loosely-held plutonium was sorbed from the solution with only a minor fraction being nascent. With only a few exceptions less than 10 per­cent of the plutonium removed from the glass at 360 days was in true solution.

The oxidation-state distributions at 250 C (Figure 2) did not differ in kind from the distributions reported in single-phase speciation experiments [1-7]. In waters that oxidized plutonium in the single-phase studies, notably the Rentfrow water, the leached plutonium in solution was principally in the upper oxidation states. Similarly, in waters that reduced plutonium, e.g. the basalt, the leached plutonium was mostly in the lower oxidation states. However, the 90 0 C data indicate a larger percentage of oxidized plutonium than in the single-phase studies. Moreover, the Pu(IV) concentration was significant in most of the glass leachates at both temperatures, whereas Pu(IV) was observed in the single-phase experiments only in ground waters con­taining appreciable concentrations of strongly complexing ligands. One pos­sible explanation is that the Pu(IV) was associated with very small colloids, small enough to pass through the 0.05-~m filter, and extractable by TTA. This possibility is reasonable in light of the overwhelming majority of part­iculate-associated plutonium. Alternatively, since all the plutonium present in the glass is in the tetravalent state (Pu02), it is possible the Pu(IV) observed may not have had time to equilibrate with the water.

The most striking result on display in these figures is the strong de­pendence of leaching behavior on ground-water composition. This dependence is summarized in Table III, which indicates that with only a single exception, all of the waters are more effective leachants than deionized water, some by a large factor. The hazard of attempting to use deionized-water leaching re­sults to predict actual ground-water behavior is obvious.

In an attempt to determine the mechanism for plutonium removal from the glass, a set of experiments was performed to monitor the change in chemical composition of ground water during glass leaching. Basalt, shale, and tuff ground waters were selected because they represent extremes in composition. The components monitored were Si02 , as a measure of the glass-matrix dissolu­tion; Na, representing release of the mobile alkali metals~ Mg, an indicator of the release of the less-mobile alkaline earth metals; Fe, because of its very high concentration in the glass, and to monitor release of multi-valent metals from the glass; and Pu.

Ground-water composition changes after 180 days' leaching are shown in Table IV. Only small changes in composition occurred at 25

0C; greater

changes at 90oC. With the exception of the tuff water, there was no signi-

106

ficant increase in the sodium-ion concentration, in marked contrast with the reported increase when deicmized water was used as leachant [11]. At 900C, significant quantities of iron were leached, expecially by the basalt water. Only modest increases in silica concentration were observed.

There were two compositional changes not caused by leaching, but by the temperature change. Except for the basalt water, the concentrations of cal­cium and magnesium decreased when the temperature was increased from 25 0 to 900

C, because of the negative temperature coefficient of solubility for sev­eral calcium and magnesium salts, notably the sulfates, causing them to pre­cipitate at 900

C. The calcium and magnesium concentrations were too low to achieve saturation in the basalt water.

To gain a better insight into the leaching mechanism, a more useful com­parison is the normalized mass loss coefficient (NL) [17], the quantity of glass which, upon total dissolution, would cause the observed increase in concentration of a given glass component in the leachant. It relates the quantity of each component released from the glass to the concentration of that component within the glass. By comparing the NL for several components it is possible to determine whether one component was being preferentially leached from the glass, or whether all components were being released in a congruent fashion, that is, strictly as a result of matrix dissolution. A large NL value indicates extensive leaching of that component. The' NL is calculated using the following equation:

~.;rhere:

VL

C. L 1.,

SA

C. 1.,g

VL·Ci,L (NL) .' = 1.,T SA.C. 1.,g

grams of glass dissolved per unit necessary t0

2produce the measured

nent 1. (gm/ cm )

= volume of leachant (L)

area in total time T concentration of compo-

= concentration of species i in the leachant (moles/L)

2 = surface area of the glass leached (cm )

concentration of species i in the glass (moles/g)

The calculated NL values for the three waters at 900C and 180 days are given in Table V. If NL '1' is the indicator for glass-matrix dissolution, sodium is removed prefer~~ti~ily from the glass except in the shale water, whereas iron and plutonium are removed at rates slower than glass-matrix dissolution. For at least these three waters, the quantity of material re­moved from the glass, be it silica, sodium, magnesium, iron, or plutonium, depends on which water is leachant. The tuff water is the most effective solvent for the glass matrix (as indicated by the largest NL value for silica) whereas the basalt water is the most effective leachant for iron and pluton­ium. Only the shale water failed to leach any of the monitored glass compo­nents appreciably. These results indicate that different ground waters leach different components preferentially. Predicting the durability of waste glasses based on leaching studies utilizing deionized water may not be conser­vative, but in fact may be optimistic. An example is the leaching of pluton-

107

ium at 90o

C, where all the ground waters were more effective 1eachants than deionized water. It is obvious that the composition of ground water contact­ing the waste glass in an actual repository may be altered by passage through the backfill material, and also by dissolution of canister corrosion products. However, these alterations would most probably make the ground water leaching behavior even more dissimilar to that of deionized water, further demonstra­ting the hazard of attempting to predict repository glass leachability from deionized-water results.

CONCLUSIONS

The quantity of plutonium leached from Battelle 80-207 glass depends on the chemical composition of the ground water used as 1eachant. The water from Grande Ronde Basalt, which contained approximately SO mg/L fluoride, was the most effective 1eachant for plutonium at 90 0 C and the water from Red Lake South well, which was quite oxidizing, was the most effective 1eachant at 2SoC. Tuff ground water was most effective at dissolving the glass matrix, but was not effective for removing plutonium from the glass, whereas shale ground water was the least effective 1eachant. In general, after 360 days greater than 90% of the plutonium removed from the glass by all of the waters was either sorbed to the glass or Teflon container walls, or was suspended in the 1eachant. The quantity of plutonium in solution, i.e. that passed through a O.OS-~m filter, was less after 360 days than it was after 180 days. Pluto­nium (IV) was more prevalent in the glass 1eachates than in experiments in which dissolved plutonium was added to these same ground waters, suggesting possible sorption of insoluble Pu(IV) onto dispersed colloidal material smal­ler than 0.05 ~m. At least in the time frame of this study, the glass was leached incongruently with sodium removed faster than the glass matrix, and plutonium and iron being removed more slowly than the matrix. Based on these results, the leaching of Battelle 80-207 glass by actual ground waters is, in general, consistent with the leaching theory developed utilizing data from deionized-water experiments. However, with only a few exceptions, all the waters were significantly more effective 1eachants for plutonium than deionized water at both temperatures.

ACKNOWLEDGEMENTS

The authors thank P. Salter of Rockwell Hanford Operations, D. Isherwood of Lawrence Livermore National Laboratory, E. Doty, K. Peter, T. Thompson J. Nuter, and R. Chappel, all members of the U.S. Geological Survey, for their assistance in collecting the various ground waters. Water analyses were per­formed by the U.S. Geological Survey's laboratory in Denver. Finally, special thanks are due to G. Bryan of Battelle Pacific Northwest Laboratory for his splendid cooperation in preparing the glass cubes used in this study.

REFERENC~S

1. Cleveland, J.M., Rees, T.F., and Nash, K.L., "Ground-Water Composition and Its Relationship to Plutonium Transport Processes,·' Plutonium Chemi­stry, W. T. Carna11 and G. R. Choppin, eds., ACS Symposium Series 216, American Chemical Society, Washington, D.C., 1983 pp. 33S-346.

2. Cleveland, J.M., Rees, T.F., and Nash, K.L., Nuc1. Techno1., ~, 298 (1983).

108

REFERENCES - Continued

3. Cleveland, J.M. , Rees, T.F. , and Nash, K.L. , Science, 221, 27.1 (1983).

4. Cleveland, J .li., Rees, T. F., and Nash, K.L. , Science, 222, 1323 (1983).

5. Rees, T.F. , Cleveland, J .M. , and (1984).

Nash, K.L. , T\Jucl. Technol. , &..~, 131

6. Cleveland, J.M., Rees, T.F., and Nash, K.L., "Americium, Neptunium and Plutonium Speciation in Ocean Water", in preparation.

7. Cleveland, J.M., Rees, T.F., and Nash, K.L, "Plutonium, Neptunium, and Americium Speciation in Selected Ground Waters", in preparation.

8. Nash, K.L. and Cleveland, J.ti., Radi~chim. Acta, 1l, 105 (1983).

9. Nash, K.L. and Cleveland, J.M., "The thermodynamics of Plutonium (IV) complexation by fluoride and its effect on plutonium (IV) speciation in natural waters," Radiochim. Acta (in press).

10. Nash, K.L. and Cleveland, J.M., "Thermodynamics of the System: americium (III) - fluoride - stability constants, enthalpies, entropies, and sol­ubility product," Radiochim. Acta (in press) .

11. Barkatt, A., Simmons J.lL and Hacedo, P.B., Nuc1. and Chem. \vaste }{g,t., 1, 3 (1981).

12. Westik, J.U. and Turcotte, R.P., :lHydrothermal glass reactions in salt brine," Scientific Basis for Nuclear Haste Hanagement, Vol 1, G. J. McCarthy, ed., Plenum Press, New York, 1978, pp. 341-344.

13. Merritt, W.F., Nucl. Techno1.,.;g, 88 (1977).

14. Clark, D.E., Zhu, B.F. and Robinson, R.S., "Preliminary report on a glass burial experiment in granite", Second International Symposium on Ceramics in Nuclear Waste Management, Chicago, Ill., Apr. 1983.

15. Pederson, L., personal communication (1983).

16. Thornhill, R.E., personal communication (1981).

17. Nuclear Waste Materials Handbook, Waste Form Test Hethods, Haterials Characterization Center, Pacific Northwest Laboratories, Richland, WA (1981) •

18. Bondietti, E.A. and Reynolds, S.A., Proceedings Actinide-Sediment Reac­tions Working Mtg., Seattle, WA, Feb. 10-11, 1976, BNlVL-2117, L.L. Ames ed., Pacific Northwest Laboratory, p 505 (1976).

19. Rees, T.F., "Special problems in the analyses for plutonium and neptun­ium in ground waters," Radioelement Analysis, W.S. Lyon, ed., Ann Arbor Science, Ann Arbor, Mich., 1980, pp. 199-205.

109

Table I. Groundwater Compositions (mg/L except pH)

Solute Basalt Tuff Shale Granite Crystal Well 5C Rentfrow Red Lake Pool South

Alkalinity (as CaC03) 146 98 530 140 256 265 80 90 Calcium <0.1 10 100 300 49 2 550 28 Iron 0.3 0.007 0.01 0.007 0.03 0.01 0.05 0.01

..... Magnesium <1 3 50 3 21 0.4 210 24 ..... 0 Manganese 0.005 0.001 0.3 0.03 <0.001 <0.001 0.008 0.084

Potassium 3.0 4.2 24 2.3 8.7 6.3 5.2 3.9 Sodium 300 50 700 300 75 140 380 280 St.rontium 0.01 0.05 3 5 1.0 . O. 018 9.9 2.5 Silica 100 70 10 10 27 57 17 15 Chloride 140 7 61 73 22 10 150 480 Fluoride 52 2.3 0.1 0.8 1.9 1.0 2.3 1.9 Phosphate 0.1 <0.01 0.02 0.02 <0.01 <0.01 <0.01 <0.01 Sulfate 75 19 2000 980 90 29 2700 23

pH 9.3 7.8 8.4 8.3 7.5 9.0 7.6 7.9

Table II. Composition of Battelle 80-207 Borosilicate Glass

Component Weight Percent

Al 203 3.65

B203 9.74

BaO 0.14

CaO 10.44

CuO 2.17

Fe203 10.12

K20 1. 97

Li 20 1. 89

MgO 3.18

Na20 7.27

P205 2.00

Si02

39.81

SrO 0.07

Ti02 4.75

ZnO 0.29

Zr02 1. 09

Pu02 0.04

Table III Total Plutonium Leached in 360 Days by Various Ground h1aters

250 C 900 C Ground Pu Leached Normalized Pu Leached Normalized Water (ECi!L) (D. 1.:::1) (ECi!L) (D.1.=1)

Basalt 69,700 12.0 77 , 200 34.2 Deionized (D. 1.) 5,830 1. 00 2,260 1.00 Tuff 3,930 0.67 6,910 3.06 Granite 6,210 1.06 24,000 10.7 Shale 8,840 1. 52 2,500 1.11 Crystal Pool 97,700 16. 7 12,700 5.63 Red Lake Well 185,000 31. 7 3,400 1. 51 Rentfrow 73,600 12.6 8,550 3. 79 Well 5C 17,300 2.96 12,000 5.30

111

Water

Basalt

Shale

Tuff

Ground Water

Basalt

Shale

Tuff

Table IV Ground l-1ater Composition (mg/L)

Before and After ISO-Day Glass Leach

Condition F Na Ca Mg Fe

25°C Before 53 260 1.5 0.07 0.190 25°C After 53 270 3.0 0.6 0.170 90°C Before 55 2S0 1.6 0.07 0.210 90°C After 55 290 7.2 2.0 4.40

25°C Before 940 50 56 <0.009 25°C After 960 51 57 0.030 90°C Before 1000 10 40 <0.009 90°C After 1000 lS 19 0.010

25°C Before 46 13 2.1 <0.009 25°C After 50 13 2.3 O.OSO 90°C Before 49 1.4 0.09 <0.009

° 90 C After 72 9.4 0.7 0.140

Table V 4 Normalized Mass Los8 (NL x 10 ) For Vari~us

Components at 90 C and lS0 Days (g/cm ) [N.C. indicates insignificant aqueous concentration change]

Na Mg Fe

7.5 lS.5 10.1 6.0

0.1 N.C. N.C. N.C.

15.1 42.6 3.2 0.2

112

Si02

110 110 120 150

14 13 0.5 0.9

65 61 60

120

Pu

0.lS0

0.006

0.009

.... .... w

80000

60000

40000

20000

0

6000

4500

3000

1.500

0

4000

3000

....1 "- 2000 ..... U O-t 1000

0

4000

3000

2000

1.000

0

10000

7500

5000

2500

0

2,50 C BASALT 90°C

L.-.-.-.. -.-.-.-.. -~ .. --]::::: r-------.. - .. -.-. (') (II

F,r-)-) (mF:: 5 (II 10

~

30 90 1BO 360 30 90 1BO 360

DEIONIZED .----------------------. 4000 i.-----------~~------~

:3000

2000

1000

o 30 90 1BO 360 30 90 180360

TUFF .----------------------,. BOOO

6000 .. -----..

4000

o ·--1iT --10--

U) 10 ....J1. "" -

2000

o 30 90 1BO 360 30 90 1BO 360

GRANITE .----------------------,.30000

15000

22500

-C\I

"7500

-6F,. L-____ ~~~----.----:g- ~

5l ~.....flIi) -o 30 90 180360 30 90 180360

SHALE

:-~~~~t--30 90 180360

DAYS

2000 ir--------------------=~--,

:1.~OO

1.000

soo

o 30 90 180360

DAYS

25°C CRYSTAL POOL 90°C 120000

80000

80000 --_._. __ ._--

....1 "-.....

30000

o

&00000

1$0000

100000

&0000

o

U 80000

O-t 60000

40000

20000

o

30000

22500

150pO

7etOO

o

30 90

30 90

30 90

180 360

RED LAKE WELL 6000

4500

3000

-rOlE I 1500

0 180360

RENTFROW 20000

15000

10000

~ I 500: I 1BO 360

NTS WELL 5C 40000

·_· __ ·_·----30000

-~---t120000

$-·-110000 o C\I

o

30 90 180 360

~O~L~ 30 90 180360

U~~~i.~ 30 90 180 360

30 90 1BO 360 DAYS

30 90 1BO 360 DAYS

~ INSOLUBLE-sORBED ~ INSOLUBLE-SUSPENDED ts:s:I SOLUBLE

Figure 1. Physical form of plutonium leached from Battelle 80-207 glass by selected ground waters.

I

.....

..... ~

kI ....:I III

3 o (f)

Ii, o ~.

'2.5°,-, 90°C BASALT ---~ 100 -.:.-, -----------_--, 100 -r------------

50 50

O tal',· -11- EilI_ E3I1W I ( i , o I 55111- sall_ ElI_ ella I if'

30 90 180360 30 90 180360

100 DEIONIZED 10°lri--:=--------------~

50 50

o o 1')- BIJ! ElI_ =U-30 90 180 360 30 90 180 360

1= TUFF 100 T' ----IT""rr"-------~

50 50

o EaV- ta'," cooI,_ ElI.'_ o 30 90 180360 30 90 180 360

100 GRANITE 100 T------------

50 50

O 5;;1'" !:iI! I , 0 , FIll. ",. EiI'b ! . , . 30 90 180360 30 90 180360

100 SHALE 100 l'I-----------~

50

O E3 - IEIII. EJIB F311_ i • j

30 90 180360 DAYS

50

o §I~'- 1§11)- ElI_ §lV--30 90 180 360

DAYS

[;r.1 ....:I III

3 o rn ~ o ~

Z50e CRYSTAL POOL 90°C lOO-r-------------------, 100-,r----------------------,

50 50

o Ell,. ai,'. r;=:[J:W 81 o t;;;;;;;IV- SI,!! &=i','! 9',"·

30 90 180360 30 90 180360

100 RED LAKE WELL l00-r'-----------------------

60 50

o I FI"I 1:311 I

30 90 180360

I L II,IL I] I,. I o 30 90 180 360

100 RENTFROW 100 lr' -----------~

50 50

o I Eb;>" !iiI!}_ Ell' ""'I'- o 30 90 180360 30 90 180 360

NTS WELL 5C l00Ir---------------------, l00-r.---------------------,

50

o I .". 9,,- MI. Mill . . \

30 90 180360 DAYS

Ii515a P1..1(lI1) IIIII1II P 1..1 (tV) _ P1..1(V+V1)

50

O BII ...

I

30 90 180 360 DAYS

Figure 2. Oxidation-state distribution of soluble plutonium leached from Battelle 80-207 glass.

THE QUANTIFICATION OF SOURCE-TERM PROFILES FROM NEAR-FIELD GEOCHEMICAL MODELS

1. G. McKi n1 ey Eidg. Institut fuer Reaktorforschung CH-5303 Wuerenlingen, Switzerland

ABSTRACT

A geochemical model of the near-field is described which quantitatively treats the processes of engineered barrier degradation, buffering of aqueous chemistry by solid phases, nuclide solubilisation and transport through the near-field and release to the far-field. The radionuclide source-terms derived from this model are compared with those from a simpler model used for repository safety analysi s.

115

I NTRODUCTI ON

As part of the safety assessment for high-level waste repositories, radionuclide source-terms to the far-field must be defined for subsequent migration modelling. In recent analyses [1, 2] rather simple near-field models have been used to derive such source-terms which are intended to be, if not realistic, at least conservative from a safety point of view. A more real i sti c approach is needed, however,' in order to assess the degree of conservatism in such simple models, evaluate the importance of particular near-field barriers in determining the source-term and identify key areas in which information is lacking. In this paper a simple but fairly comprehensive model of the Swiss near-field is ,described which allows quantitative evaluation of engineered barrier lifetimes, gross variations in aqueous chemistry in various regions of the near-field, nuclide solubilisation and transport to the far field [3]. Resultant release profiles are compared with those derived from a simple conservative model.

DESCRIPTION OF THE SWISS NEAR-FIELD

The reference Swiss disposal scheme involves emplacement of vitrified high-level waste, encapsulated in cast iron (or steen, in a granitic host rock at a depth of about 1300 m. Waste containers are positioned horizontally in the centre of 3.7 m diameter drifts which are backfilled with compressed bentonite (Fig. 1). For safety analysis, a repository containing 5895 waste canister is considered. The material inventory in the engineered barriers surrounding each canister is summarised in table 1. The near-field also includes the mechanically damaged rock around the tunnel which is considered to extend about 4 m beyond the drift wall.

Table I

Material inventory in the near-field (per canister)

Materi al

Glass Steel fabrication container Fabrication void Canister Backfill (water saturated)

0.15 0.01 0.03 0.9

52.8

116

Mass (kg)

405 75

- 3 6.5x10

5 1.1x10

/

Backfill (Bentonite)

~~~""""'~""""'~"S~!Io.----7'L--r-- Canister (Fe)

Wasteform (Glass)

5.00 (pitch)

//

Fig. 1: Reference Swiss disposal geometry (dimensions in metres)

NEAR-FIELD MODELS

The conservative model used to derive source-terms for migration modelling assumes that all waste canisters fail 1000 years after emplacement. The waste glass matrix then dissolves congruently at a rate dependent only on the exposed surface area (1(f7 g/cm 2/day for the base case). A simpl e computer code evaluates this release taking into account variation in surface area as dissolution progresses, radioactive decay/ingrowth and possible solubility constraints [4J. The solubility calculations assume that the entire water flux through the repository is 'available ' for saturation and that effective solubility for multiple-isotope elements is proportional to their relative isotopic abundance in the source at that time. Input solubility limits for the actinides and Tc are calculated for far-field conditions by the thermodynamic equilibrium code MINEQL/EIR [5J while values for other elements are taken from the literature.

A more realistic model of the near-field first requires an evaluation of possible reactions between groundwater and repository materials which allows quantification of engineered barrier longevity and gross variations in water chemistry between different zones in this region. Solubility limits can then

117

be estimated for near-field conditions. These solubilities can be combined with the previously described glass dissolution model and a more realistically evaluated 'available ' groundwater flux to derive radionuclide input functions at the inside of the bentonite backfill. A very simple diffusive transport calculation then allows source-term functions outside the bentonite to be derived. This treatment explicitly includes examination of possible perturbing processes such as radiolysis and colloid formation during leaching or at redox fronts.

GEOCHEMICAL REACTIONS IN THE NEAR-FIELD

The materials used for the engineered near-field barriers are thermodynamically unstable in the presence of host-rock groundwater and resultant reactions will alter the chemistry of both solid and liquid phases.

Reaction of bentonite with groundwater would involve ion-exchange of solution cations for Na+ (as sodium montmorillonite is the main mineral component of the reference bentonite), solution pH would increase to a value in the range 7 - 8.5 with an associated decrease in redox potential [3, 6]. Some illitisation of the montmorillonite would also occur but, even if the very slow kinetics of such reaction were ignored, the low rate of 6Upply of K+ would limit reaction to ~ 2% of the total buffer in a period of 10 y. Both such stoichiometric calculations and observation of natural analogues [7] justify the further assumption that constant chemical agd hydrologic conditions prevail within the backfill for periods > 10 years for the base-case scenario.

Based on an empirical corrosion rate of 50 p/yr, the iron canister is assumed to fail mechanically after about 1000 years but this calculation is very conservative in the extrapolation of laboratory corrosion-rates and pitting measurements to the long timescale involved. Taking into account the additional decrease in corrosion kinetics resulting from the more alkaline conditions realistically expected in the near-field and possible physical or thermodynamic inhibition due to build-up of corrosion product layers or any hydrogen produced by anoxic corrosion [3], corrosion rates at least an order of magnitude lower than the reference value may be realistically expected. Even after canister failure remnant iron and oxy-hydroxide corrosion products would be oxidised only very slowly in the near-field environment and total cogversion to a fully oxidised form (eg. Fe203) would be expected to take> 10 y [7]. In addition, the deep granitic groundwater is actually oversaturated in Fe 2+ at the higher pH of the near-field and hence continuous, slow precipitation of Fe(OH)2 (or FeSi04) would be expected in this region.

The differences between Eh/pH conditions found in the near-field and those expected on the basis of the geochemical reactions above are illustrated in table 2. Limiting elemental solubilities of some species calculated by MINEQL/EIR are also presented in this table [3, 5] from which it can be seen that a quite significant decrease is observed in near-field conditions for the case of Tc. Although the general applicability of such thermodynamic codes, and even of the entire Eh concept, as applied to low temperature groundwaters has been questioned (eg. [8]), the porous canister corrosion products would be expected to provide an almost ideal redox buffer and thus agreement between

118

'canister-zone ' solubilities and far-field values used in the conservative model may provide an important justification for the latter. Species which may be more soluble in a near-field environment (eg. Pu) are thus identified as worthy of more detailed study if the conservative model assumes solubility limited release. It should be noted that, for many relevant fission product elements, solubility data are even more sparse than for the actinides and the literature values used for modelling are order of magnitude estimates which do not justify analysis of the type above.

Tabl e II

Solubility limits

pH EO * Log Solubility ( M) Tc Th U Np Pu Am

* Redox nb

cQnditions EO = 0 EO = 220 EO = 407

in the near- and far-fields

Canister Bentoni te Far-fi el d 7 - 8.S 7 - 8.S 6.8 o - 220 220 - 407 220 - 407

-8 -4. S -4.3 -8 -8 -7.8 -9 -9 -8.6 -8 -S -8.7 -3 -6. S -7.0 -4.7 -4.7 -4.3

assumed to be given by Eh = EO -S9 pH (mV) iron/magnetite or H2/H+ equilibria haematite/magnetite equilibrium ferric hydroxide/siderite equilibrium.

HYDROLOGIC CONSTRAINTS

The conservative model pessimistically assumes that released radionuclides may saturate the entire water flux flowing through the repository and ignores retardation in the bentonite backfill. Rad;onuclides may then be classified as either solubility limited or matrix dissolution limited depending on which process constrains their release to the far-field.

Current hydrologic models indicate that, for the reference repository, more than 90% of the total water flux occurs in discrete 'kakirite ' zones which can be identified and avoided during emplacement. In addition, of the water flowing between emplacement drifts only that within 2-4 m (accounting for the mechanically damaged zone) can be reasonably expected to be available for nuclide uptake [1, 3J. It is thus reasonable to assume that the available flux for nuclide uptake is 2 orders of magnitude lower than the total flux and still conservatively neglect the further reduction in 'available flux ' caused by transfer resistances at the glass/canister, canister/bentonite and bentonite/rock interfaces.

119

Solute transport through the bentonite backfill will occur predominantly by diffusion. From experimental and theoretical studies, the pore structure of compressed bentonite is expected to preclude transport of colloids or suspended particulates [ego 9J and, 6based on previous arguements, this situation is expected to persist for> 10 y. Solute retardation in such a medium is poorly understood at present but, for semi-quantitative calculations, it should be reasonable to assume a constant distribution coefficient (Kd), reversible sorption model [10J. From a simple, infinite porous medium calculation [3J, times calculated for breakthrough of 1% of the inner surface concentration at the tunnel wall range from about 1000 years for poorly retarded anionic species (eg. I, Kd = 0.005 ~/k9)JO about lOG y for strongly sorbed species (eg. Zr, Pu, Am, Th etc., Kd = 5 /kg).

The effect of these two processes can be cl early seen if the output erofil e for the 14n+11 chain derived from the conservative model (Fig. 2a - [4J) is compared to that from a semi-quantitative analysis of this realistic model (Fig. 2b). Retardation delays output to such an extent that some unsupported nuclides decay completely within the bentonite (Cm-245, Am-241). For long-lived species (e.g. Np-237) this delay has lit61e effect on their output concentrations - although in cases where delays ~ 10 years occur the significance of further calculations is debateable. In intermediate cases, excess releases of relatively short-lived daughters decay during transit and hence source concentrations to the far-field are at secular equilibrium with long-lived parents (e.g. U-233, Th-229). The nuclide release rate, whenever it occurs, is directly affected by the water fl ux only for sol ubil ity 1 imited species (and their supported daughters) when it is directly proportional to thi s parameter.

120

x 3 LL

W C/) <l W ....J ILl a:::

b)

a::: <l w >-....... C/) W ..J 0 ~

X

:3 LL

ILl C/) <l ILl ....J ILl a:::

\ .-.. - .. _ .. _ .. ---. .,,>...-.. "--"-"-'\ _____ :=.t __ -- --------------------"'t ./. """""', .-.-._ 1

'\ .-- .~ I --'--' " ' -"--' .\ ~ .~

CM-245 AM-241 NP-237

U - 233 TH - 229

II ;.,

10-13~ -----------....--L----------,..---111.---1d

10-1

10.5

10.9

10.13

103

1d TIME (y)

./ "" .--- ....... ~---------'i

/ -'- "' / .,..,.' 1/ /./;" .. ----.. - .. - ...

1(// I I . ,

107

TIME (y)

Fig. 2: Radionuclide release rates to the far-field from the entire repository: a) conservative, b) realistic model.

121

RADIOLYSIS

Although it will not be discussed in detail, the consequences of the radiolysis of water have been considered in evaluating realistic geochemistry. From the assumptions that 1) radiolysis occurs predominantly due to alpha de~ay, 2) all the energy from 0.3% of all alpha decays in the period 105 - loP Y causes radiolysis and 3 ) that all 'reductant' (H2) produced is lost to the system without recombination, it was calculated that about 450 moles/can~ster of net oxidant (represented as H202) could be produced in the first 10 y [3J. Although such radiolysis might cause more oxidising conditions to prevail in the micro-environment of the glass surface, it could be easily buffered by the canister corrosion products (eg. if the entire canister were initially corroded to Fe304, this 'radiolytic oxidant' could be consumed by conversion of < 2% of the canister mass to Fe203). Any redox front formed (eg. [llJ) and associated colloidal precipitates would thus be contained within the bentonite barrier. If a redox gradient exists between the glass surface and the buffering corrosion products, net solubility will be determined by the lowest value encountered which could be considerably lower than the 'canister' value in some cases (eg. Pu).

CONCLUSIONS

A simple geochemical analysis of the near-field for the reference Swiss HLW disposal scheme indicates that a chemical environment (reducing, slightly alkaline) will be established in this region which, together with the low availability of water, will limit the release rates of many radionuclides. The physical properties of the backfill - preventing colloid transport and severely retarding diffusion of many dissolved species - will result in many nuclides decaying completely within this region and the release of others being delayed for very long periods. Both these hydrologic6and chemical properties would appear to be stable over a timescale of about 10 y and any radiolytic effects would be contained within the canister corrosion products and have no influence on far-field release functions.

ACKNOWLEDGEMENTS

The work reported in this paper was greatly aided by the encouragement and constructive criticism of J. Hadermann. The • conservative' model was developed by R. Hartley and MINEQL/EIR calculations were performed by M. Schweingruber. H.P. Alder, H. Flury and C. McCombie are thanked for their interest and Nagra for financial support.

122

REFERENCES

[1] Nagra: "Endl ager fuer hochaktive Abfaell e: Das System der Sicherheitsbarrieren". NGB 85-04, Nagra, Baden, in press.

[2] KBS: "Fi nal storage of spent nucl ear fuel - I II Barriers". SKBF /KBS, Stockholm, 1983.

[3] McKinley, loG.: liThe geochemistry of the near-field." NTB 84-48, Nagra, Baden, in press.

[4] Hartley, R.: "Description of the source-term model for far-field migration calculations". Manuscript in preparation.

[5] Schweingruber, M.: "Actinide and Tc solubilities in deep-crystalline reference groundwater defined for Nagra Project 'Gewaehr'll. TM-45-84-30, EIR, Wuerenlingen, 1984.

[6] Snellman, M.: "Chemical conditions in a repository for spent fuel il•

YJT-84-08, YJT, Helsinki, 1984.

[7] Chapman, N.A., McKinley, I.G., Smell ie, J.A.T.: liThe potential of natural analogues in assessing systems for deep disposal of high-level radioactive waste". NTB 84-41, Nagra, Baden, 1984.

[8] Lindberg, R.D., Runnells, D.O.: "Ground water redox reactions: An analysis of equilibrium state applied to Eh measurements and geochemical modelling". Science, 225, 925-927,1984.

[9] Pusch, R.: "Stability of bentonite gels in crystalline rock-physical aspects. II TR-83 -04, KBS, Stockholm, 1983.

[10] McKinley, I.G~, Hadermann, J.: "Radionucl ide sorption database for Swi ss safety assessment. II NTB 84-40, Nagra, Baden, 1984.

123-124

A CONCEPTUAL MODEL FOR REPOSITORY SOURCE-TERM EVALUATION

M. J. Apted, D. H. Alexander,* A. M. Liebetrau, A. E. Van Luik,** R. E. Williford and P. G. Doctor, Pacific Northwest Laboratory, P. O. Box 999, Richland, Washington; *U. S. Department of Energy, Washington, D. C.; **Weston Consultants, Rockville, Maryland.

ABSTRACT

A new source-term model is described that will yield probabilistically determined estimates of the times-to-penetration by various corrosion mechanisms for metallic barriers of waste packages. After penetration, the model evaluates radionuclide release rates as a function of groundwater hydrology, dissolution/precipitation reactions, sorption processes, inventory changes, and variations in the geochemical/radiation environment. Integrated repository releases will be calculated over time by summing the probabilistically generated releases from the individual waste packages.

125

INTRODUCTION

The U. S. Department of Energy (DOE) is developing a performance assessment strategy to demonstrate compliance with the standards and technical requirements for the safe isolation of high-level nuclear waste (HLW) set by the U. S. Environmental Protection Agency (EPA) and the U. S. Nuclear Regulatory Commission (NRC). The post-closure portion of the performance assessment strategy requires coordination between site-specific research and model development, and expert technical judgment on expected and disruptiv' scenarios.

Post-closure performance assessment of the geologic repository system can be divided into a Source Term Model and a Contaminant Transport Model (Figure 1). The Source Term Model would be used to estimate the expected range for canister lifetime and containment radionuclide release rates from the population of waste packages, as well as provide input to the Contaminant Transport Model. The Contaminant Transport Model would then estimate cumulative releases to the accessible environment.

This paper focuses on the structure and components of one approach to a Source Term Model. In addition, it identifies processes that should be considered for assessment of the performance of the canister, cladding, and spent fuel components of the engineered barrier system. A companion paper [lJ

SIMULATiON OF SYSTEMS AND SUBSYSTEMS

EXPERT JUDGMENT ON NORMATIVE AND DISRUPTIVE SCENARIOS

FIELD AND LABORATORY RESEARCH

Figure 1. Components of Geologic Repository System Post-Closure Performance Assessment Strategy (after Lyon, 1982)

126

Figure 2. Schematic of Waste Package Barrier Penetration and Radionuclide Release

addresses the need for a broad, inclusive approach to performance assessment and the potential ramifications of such an approach. This Source Term Model (Figure 1) would provide the means to assess compliance with NRC performance criteria for waste package lifetime and engineered barrier system release rate [2J. It is recognized that limitations in data and models may require that such a model be used in conjunction with conservative bounding assumptions on a site- and design-specific basis.

As presently conceived, this model will evaluate radionuclide releases from a realistic waste package/repository configuration, avoiding the physically impossible scenarios that are often called "worst case." There are two components to the Source Term Model. The Barrier Penetration Model simulates water penetration of the container and the Zircaloy cladding before reaching the spent fuel (Figure 2, tl-t2). After penetration of the metallic barriers, the release of waste to the water and migration outward through corroded openings (Figure 2, t3) will be calculated by a Release Model. Release of radionuclides from the waste package will be controlled by the limited exposed surface area of the spent fuel as well as chemical interactions with primary and corrosion products that will line exit paths. By accounting for these processes and events, a Source Term Model can be used to establish a more realistic failure scenario. Results from this model can also be used to guide the future testing and design modifications in order to reduce uncertainty in performance of the engineered barrier system.

127

The Source Term Model proposed here differs from several existing waste package performance assessment models such as BARRIER [3J and WAPPA [4J in that it will stochastically estimate the distribution of canister failures and radionuclide releases over time from the total population of waste packages. Similar probabilistically-based approaches have been adopted by the Canadian [5,6J and Swedish [7J nuclear waste repository programs, and proposed for at . least one U. S. program [8J.

BARRIER PENETRATION MODEL

Model Description

The Barrier Penetration Model is designed to estimate the time-to­penetration of a typical waste package under repository conditions in as realistic a manner as the current state -of knowledge allows. For modeling purposes, a barrier is defined as "penetrated" at the time that its net thickness becomes zero at some locale. The barrier penetration event is preceded by a "damage" period and followed by an aperture "growth" period. "Damage" is defined as a reduction in the local net thickness of a barrier via corrosion or fracture propagation. Aperture growth is defined as the subsequent (post-penetration) change in the cross-sectional area of local penetrated regions. Note that aperture growth can mean an increase in aperture size or numbers (as in corrosion processes) or a decrease, as in the plugging of a small fracture by corrosion products.

The corrosion of barriers is linked sequentially in a time-dependent fashion. The time at which the first barrier is penetrated initiates corrosion of the next barrier. In addition, the progress of corrosion for all barriers is tracked by an aperture growth model.

Mode 1 Structure

The model is composed of a time-linked sequence of modules, a structure that allows for a widely ranging mix of stochastic and deterministic components and one that is easily modified to incorporate current information. The model is driven by a set of input information that describes the canister, its contents, and the repository environment.

Figure 3 reveals in a general way the modular structure and basic logic of the Barrier Penetration Model. It contains modules that describe the steam corrosion mechanism and penetration models for canister (low-carbon steel in this example) and Zircaloy (Zry) barriers. The modules are ordered in time. The specific model in each depends upon a number of factors, an important one of which is the host medium (basalt, salt, tuff, granite).

The model will perform as follows. Parameters are chosen to describe the repository design, the canister and its contents, and initial environmental and ambient conditions [5,8J. Appropriate probability distributions will be assigned to describe the ranges and variability of initial values of these parameters. A canister is then selected randomly from the "population" of canisters defined by these initial distributions.

Figure 4 shows detail for a possible model of the time-to-penetration of a low-carbon steel (LCS) barrier. Parameters and variables that change spatially and temporally are updated as necessary. The updated values become

128

Repository Description 1, Design Parameters 2. Site Characterization

Canister Description

Time

Steam Corrosion

Repeat for Each Canister

Penetration Model

(LCS Barrier)*

Penetration Model

(Zry Barrier)

t---r-~ Release Model

LCS Aperture Zry Aperture Growth Growth Models Models

Figure 3. Flow Chart for Barrier Penetration Model

inputs to the LCS corrosion model. Corrosion of the LCS barrier is caused by a number of interdependent, time-dependent physical degradation processes such as uniform corrosion, stress corrosion, pitting, and fracture. All degradation processes are simulated through time until at least one of them results in penetration. A decision as to whether to continue simulation of LCS corrosion can be made at that time. The decision depends largely upon the nature of the model for aperture change (discussed below). The overall model can incorporate different corrosion mechanisms for Zircaloy than for LCS.

A specific example of a corrosion (penetration) rate model is that of Westerman and others [9J, who present a model for uniform corrosion of LCS in a salt environment. The authors used laboratory data to develop an equation that relates uniform penetration rate to temperature, radiation dose rate, flow rate, and other parameters. The parameters in such a model can be assigned uncertainties on the basis of best available information from site characterization and design studies. Deterministic models will be used, where available, in connection with estimates of parameter uncertainty. Where suitable mechanistic models are unavailable, probability models (distributions) will be used instead.

Before the Barrier Penetration Model can be implemented, aperture growth models must be developed. For the various barriers, these models specify the nature of corrosion products and how the corroded surface area changes with time; this directly affects the release of radionuclides. A "worst case" model would specify that the area of the failed surface becomes maximal at the time of penetration. For this model, there is no need to track corrosion mechanisms for additional time. A more realistic model would specify the rate of aperture change with time by taking account of the effects of important corrosion mechanism and products. In this case, it is worthwhile to continue simulation of LCS corrosion after penetration of the LCS barrier because the cumulative effect of various corrosion mechanisms could be significant.

129

Discription of Repository and Canister

" Distance Attenuation Models,

Time-Change Models

Flaw Size and Flaw Frequency

.. ..

Distributions

" • • • • " Uniform Stress Pitting Fracture

Corrosion Corrosion Corrosion Model: Model:* Model: T23 T24

T21 T22

• -y- t j

• T2 = T1 + min {T21 , T22, T23, T24}\

T1 ~ ... 2

Figure 4. Detail of Corrosion Mechanisms including including Barrier Penetration Model

..

The model, as presently formulated, will simulate one canister at a time until a repository is developed. This procedure is based on the assumption that the waste packages behave independently; that is, that the penetration of one canister does not affect the penetration times of its neighbors. If this simplifying assumption is determined to be unrealistic, it will be necessary to simultaneously simulate the times-to-penetration of all the canisters in the repository.

The Barrier Penetration Model takes into account not only spatial and temporal variability in geochemical parameters and canister descriptors, but also uncertainties in their initial values. Insofar as possible, parameter values and their uncertainties are estimated from actual data. The alternative is to systematically postulate an array of plausible scenarios and examine their consequences. Models that describe spati.al and temporal changes depend, of course, upon the host material and descriptors of both repository and canister design.

Mode 1 Inputs

The Barrier Penetration Model will be driven by a number of parameters. Model inputs consist of the set of initial values and uncertainties for these parameters [5,8J. Where appropriate, initial values will be selected from probability distributions that describe the range and variability of the input parameters. The input parameters fall naturally into three classes:

130

environmental and/or geochemical parameters, repository design parameters, and parameters that describe the canister and its contents.

Environmental and geochemical parameters include the following: host material (e.g., basalt, salt, tuff, granite), temperature, pressure, solid phases, pH, Eh, radiation fields, and groundwater flow descriptors (rates, direction, etc.). Repository design parameters include the following: number of canisters, number and type of engineered barriers, and time of closure. Canister descriptors include the following: dimensions, composition (e.g., LCS) and thickness, waste type (spent fuel, commercial high-level waste), initial radiation, dose rate, initial temperature at centerline, and the probability that a canister contains pre-damaged fuel rods.

Model Outputs

The basic output of the Barrier Penetration Model is a series of stochastically determined times-to-penetration. This output can be described as a point process of canister penetration times. This process depends upon the values established for model input parameters.

In addition, output is structured to understand the factors that affect penetration times and, more importantly, to drive the Release Model. Output for the first purpose includes a breakdown by barrier and corrosion mechanism of the times-to-penetration for the simulated canister. The values of parameters at the time of penetration, which are needed to estimate leach rates and chemical reaction times, are outputs for the second purpose. These parameters include temperature, redox conditions, pH, Eh, F-, Cl-, and S-2. Unless aperture growth occurs over a short period of time, a necessary output from the Barrier Penetration Model will be information about aperture growth (composition and growth rates of alteration phases, for example) for use in the Release Model.

The point process of canister penetrations can be used to estimate the probability of canister penetration at any specified time and to give an associated estimate of uncertainty. Such information is crucial to assessing compliance with NRC regulations regarding waste package (canister) lifetime [2J. Moreover, these times can be used in connection with the Release Model to estimate total repository release.

RELEASE MODEL

Model Description

Predicting the rate of radionuclide release subsequent to the loss of physical containment is a vital part of the overall performance assessment of geologic repositories. The purpose of the Release Model is to calculate expected radionuclide releases and subsequent migration for a range of different repository conditions. This model will use the output of the Barrier Penetration Model and additional site-specific data to evaluate conformance to both the engineered barrier release rate limit set by NRC and the EPA limits on radionuclide release from the repository to the accessible environment.

The Release Model within our Source Term Model will be coupled in two parts (Figure 5). The first part represents release and uses a generalized

131

mass transport theory [10,11J, combined with chemical mass balance and mass action constraints within the engineered barrier system, to calculate radionuclide release rates. This first part will be initiated by the output parameters from the Barrier Penetration Model. These include initial environmental conditions, exposed surface area for the waste form from corrosion-aperture growth calculations, and the starting time for waste form/water interaction. The second part, representing migration, will apply subsequent chemical and physical processes that may retard or attenuate the migration of radionuclides within the engineered barrier system. The following sections address the structure of the Release Model and describe the necessary inputs to as well as the desired output from such a model.

Model Structure (Release and Migration)

Previous models for radionuclide release have been published on both a repository-specific [12,13J and generic basis [10,14J. A common aspect of most of these approaches is the use of mass transport theory. This theory has recently been considerably expanded so that exact analytical solutions for waste package performance assessment now exist [10J. Use of such analytical mass transport solutions provides for simple and accurate calculations of

Function of Contrasting

Hydraulic Conductivities

Figure 5. Schematic of Processes Affecting Release Model of Repository Source-Term Model (after Lyon, 1982)

132

release rates from waste/barrier/rock systems. In addition, this approach can be used to parametrically determine the limits of validity for the model and sensitivity of the results to design and environmental factors. The relevance of mass transport models to the extremely different conditions of proposed repositories (e.g., saturated versus unsaturated groundwater flow, groundwater flow versus brine migration) has recently been addressed, but is not yet reso 1 ved [15J.

There remain several modifications to present mass transport models that are needed to simulate important aspects of the repository setting. The effect of the extremely limited surface area of the waste form exposed directly to groundwater must be included, and the nature of flow in and around such a penetrated canister should be estimated. The application of saturated flow models to conditions of two-phase flow needs to be addressed. Further review of the approach equating fracture flow to equivalent porous flow is also required. In addition, much of the data on waste form/groundwater interaction does not accurately represent actual repository conditions [10J and are, therefore, of· questionable use in mass transport modeling. These limitations notwithstanding, release rate calculations based on mass transport calculations provide a technically defensible and programmatically achievable approach to the assessment of a repository source term.

Upon release from the waste form surface, radionuclides will migrate away by fluid flow or solute transport. Within a homogeneous porous packing (backfill/buffer) material or a water-filled void space around a waste canister, the migration of radionuclides can be expressed by the exact analytical equations previously developed from mass transport theory. Transport within the engineered barrier system may be retarded or attenuated by chemical or physical processes. Mass transport equations modified by several such processes, including radioactive decay and sorption, have been derived for radionuclide migration [10,l1J.

The effect of reversible sorption on radionuclide migration has been extensively studied [16-l9J. Over the relatively short pathways in the engineered barrier system, sorption does not significantly mitigate the steady-state release rate of radionuclides [10J. Exceptions occur for radionuclides with short half-lives (e.g., cesium-137, strontium-90) or with extremely high sorption coefficients. Irreversible sorption (co-precipitation), in which sorbed radionuclides could become incorporated within the sorbing phase (including corrosion products), may also be an important process [18,19J.

Colloid formation and transport represents a potentially significant eerturbation to mass transport models of radionuclide release and migration L20,21J. If colloidal particles form and are not filtered, they may transport additional radionuclides at the water velocity. Studies of colloid transport through clay and sand packing/backfill materials have demonstrated that physical filtration and aggregation effectively prevent the migration of radionuclide-bearing colloids [20,21J. Furthermore, the general reports of radionuclide concentrations in waste form IIleach ll solutions filtered through an approximately 400 nm filter undoubtedly includes a large colloidal fraction [22J. Release rates based on such test results would, therefore, overestimate the amount of radionuclides that are truly soluble and transportable. Empirical corrections to radionuclide migration rates arising from colloid transport will be applied, as necessary, on a site-specific basis.

133

Current radionuclide migration models require improvements in several areas. Relevant waste package and geochemical reactions are typically observed to be at nonequilibrium (or steady-state) conditions, although they are generally treated as controlled by equilibrium solubility constraints [23J. Directly coupled or iteratively coupled migration models linking chemical reactions and fluid flow [24J have been applied to far-field waste contaminant transport, but the data base and code development for application to radionuclide migration are still lacking. Conversely, there may not be a need to incorporate such coupling over the limited dimensions of the engineering barrier system. Coupling of radiolytic reaction rates with geochemical reaction rates may be needed to evaluate potential pH and redox effects on radionuclide solubiltiy and transport.

Model Inputs

The initial environmental conditions used in the Release Model are derived from the output of the Barrier Penetration Model. These correspond to the temperature, pressure, radiation field, groundwater chemistry, etc., at the time of first loss of containment. The results of the aperture growth calculation are also used in the Release Model to determine available waste form surface area exposed to water (or steam), the composition of corrosion products, and the changes in the corrosion aperture with time.

In addition, specific data on properties of waste forms, individual radionuclides, and hydrodynamic parameters are required. These data include: identity of solubility-limiting phases, dissolution rate of waste form matrix at long time periods, sorption isotherms for radionuclides on fresh and altered engineered barrier materials, waste form inventories of radionuclides 1000 years after emplacement, convective flow direction and rate, aqueous diffusion coefficients of radionuclides, and the porosity (or distribution of fractures) in engineered barrier systems.

Model Output

The Release Model will provide an estimate of radionuclide release rates for the overall repository system as a function of time. Integrated engineering system releases for all radionuclides are calculated by summing releases from the individual waste packages that have lost containment (Barrier Penetration Model) and are undergoing slow radionuclide release (Release Model).

The output from this model can be structured to show the progressive effects on the release rate by successively including the distribution of canister failures, time-dependent exposure of waste form surface area, geochemical interaction effects, and radioactive decay in the model. By considering the effects of these separate factors on radionuclide release, it should be possible to evaluate their relative importance and, therefore, to help guide future research and testing.

SUMMARY

As part of DOE's overall HLW repository postclosure performance assessment strategy (Figure 1), a Repository Source Term Model is being developed [1J. This model evaluates engineered barrier system performance stochastically by sampling from ranges of input parameters that vary as a

134

function of time and of repository population variability. The model simulates canister penetration and aperture growth as a function of time (Figure 5). After canister penetration, Zircaloy corrosion is modeled, estimating cladding penetration and aperture growth as well as the resulting exposed surface area of spent fuel to groundwater. The model simulates the release of radionuclides from spent fuel using mass transport theory and appropriate site-specific constraints. Physico-chemical interactions between migrating radionuclides and engineered barrier materials, including corrosion products of iron and Zircaloy, are also explicitly included in the analysis.

This new model will be used to assess radionuclide releases to the geosphere as a function of the probabilistically determined failures in population of waste packages. The input and overall structure of this model will be flexible to accommodate site-specific features of engineered barrier systems. The model will also permit assessment of compliance with the NRC requirements on waste package lifetime and engineered barrier system release rate, as well as evaluation of importance to safety of system components.

REFERENCES

1. Alexander, D. H., et al.: "Conceptual Model for Deriving the Repository Source Term," in Scientific Basis for Nuclear Waste Management VIII, eds. C. Jantzen, J. Stone and R. Ewing. Materials Research Society Proceedings, 1985.

2. U. S. Nuclear Regulatory Commission: "Disposal of High-Level Radioactive Wastes in Geologic Repositories," 10 CFR 60, Federal Register 48 (120), 28194-28229 (1983). --

3. Lester, D. H., et al.: "System Study on Engineered Barriers: Task 3 -Barrier Performance Analysis." ONWI-2l1, Science Applications, Inc., for Battelle Memorial Institute, Office of Nuclear Waste Isolations, Columbus, Ohio, 1982.

4. INTERA Environmental Consultants, Inc.: "WAPPA: A Waste Package Performance Assessment Code." ONWI-452, Battelle Memorial Institute, Office of Nuclear Waste Isolation, Columbus, Ohio, 1983.

5. Lyon, R. B.: "Nuclear Waste Disposal: The Interface Between Performance Assessment and Research," in Scientific Basis for Nuclear Waste Management, ed. S. V. Topp., Vol. 16. Elsevier Science Publishing Co., New York, NY., 1982.

6. Dormouth, K. W. and R. D. Quick: "Accounting for Parameter Variation in Risk Assessments for a Canadian Nuclear Fuel Waste Disposal Vault," International Journal of Energy Systems, 1, p. 125, (1981).

7. SKBF, Swedish Nuclear Fuel Supply Company: "Final Storage of Spent Nuclear Fuel," Vols. 3 and 4. KBS-3, S-102 48, Stockholm, Sweden, 1983.

8. Sagar, B., P. W. Eslinger and R. G. Baca: "Probabilistic Modeling of Radionuclide Release at the Waste Packages Engineered Boundary of a Repository in Basalts." Draft RHO-BW-ST-68P, Rockwell Hanford Operations, Richland, Washington, 1984.

9. Westerman, R. E., J. L. Nelson, S. G. Pitman, W. L. Kuhn, S. J. Basham and D. P. Moak: "Evaluation of Iron-Base Materials for Waste Package Containers in a Salt Repository," in Scientific Basis for Nuclear Waste Management VII, ed. G. L. McVay, pp. 427-436. North-Holland, New York, 1984.

10. Chambre, P. L. and T. H. Pigford: "Prediction of Waste Performance in a GeologiC Repository," in Scientific Basis for Nuclear Waste Management VII, ed. G. L. McVay, pp. 985-1008. North-Holland, New York, 1984.

135

11.

12.

13.

14.

15.

16.

17.

18.

19.

20.

21.

22.

23.

24.

Zavoshy, S. J., P. L. Chambre and T. H. Pigford: "Waste Package Mass Transfer Rates in a Geologic Repository," in Scientific Basis for Nuclear Waste Management VIII, eds. C. Jantzen, J. Stone and R. Ewing. Materials Research Society Proceeding, 1985. Wood, M. 1., J. F. Relyea, J. Myers and M. J. Apted: liThe Near-Field Waste Package Environment in Basalt and Its Effect on Waste Form Releases," CONF-831217, Proc. 1983 Civilian Radioactive Waste Management Information Meeting, 1983. Neretn i eks, 1. and A. Rasmuson: "An Integrated Approach to the Description of Radionuclide Release and Transport in the Geosphere," in Scientific Basis for Nuclear Waste Management VII, ed. G. L. McVay, pp. 269-278. North-Holland, New York, 1984. Montrose, C. J., P. B. Macedo and A. Barkatt: "Phenomenological Models of Nuclear Waste Glass Leaching," Chapter 6 in Final Report of the Defense High-Level Waste Leaching Mechanisms Program (J. E. Mendel, compiler). PNL-5157, Pacific Northwest Laboratory, Richland, Washington, 1984. Chambre, P. L. and T. H. Pigford: "Comments on: Review of Mass Transfer Theory in Waste Package Modeling." UCB-NE-4050 (and LBS-18223), University of California, Berkeley, California, 1984. McKinley, 1. G.: "Techniques for Applying Complex Sorption Data from Mechanistic Studies to Transport Models," in The Geochemistry of High-Level Waste Disposal in Granitic Rocks, eds. N. A. Chapman and E. P. Sargent. AECL-8361 (also EUR-9196), Atomic Energy of Canada Limited, Pinawa, Manitoba, Canada, 1984. Couture, R. A. and M. G. Seitz: "Sorption of Anions of Iodine by Iron Oxides and Kaolinite," Nuclear and Chemical Waste Management, 4, 301-306 (1983). Barney, G. S.: "Radionuclide Sorption and Desorption Reactions," in Geochemical Behavior of Disposed Radioactive Waste, eds., G. S. Barney, J. D. Navratil and W. W. Schulz, Amer. Chern. Soc. Symp. Sere 246, pp. 3-23 (1984). Meyer, R. E., W. D. Arnold and F. 1. Case: "Valence Effects on the Sorption of Nuclides on Rocks and Minerals," NUREG/CR-3389, ORNL-5978, Oak Ridge National Laboratory, Oak Ridge, Tennessee, 1984. Champ, D. R., W. F. Merritt and J. L. Young: "Potential for the Rapid Transport of Plutonium in Groundwater as Demonstrated by Core Column Studies," in Scientific Basis for Nuclear Waste Management V, ed. W. Lutze, pp. 745-754. North-Holland, New York, 1983. Avogadro, A. and G. DeMarsily: "The Role of Colloids in Nuclear Waste Disposal," in Scientific Basis for Nuclear Waste Management VII, ed. G. L. McVay. North-Holland, New York, 1984. Apted, M. J., G. L. McVay and J. W. Wald: "Dissolution of Specific Radionuclides: Actinide Elements," in Final Report of the Defense High-Level Waste Leaching Mechanisms Program, pp. 4.20-4.49. PNL-5157, Pacific Northwest Laboratory, Richland, Washington, 1984. Grambow, B.: "A General Rate Equation for Nuclear Waste Glass Corrosion,1I in Scientific Basis for Nuclear Waste Management VIII, eds. C. Jantzen, J. Stone and R. Ewing. Materials Research Society Proceedings, 1985. Pearson, F. J. Jr.: "Linking Flow and Geochemical models: Status and Problems of Developing a Fully-Coupled Model ,II in The Geochemistry of High-Level Waste Disposal in Granitic Rocks, eds. N. A. Chapman and F. P. Sargent. AECL-8361 (and EUR-9196), Atomic Energy of Canada Limited, Pinawa, Manitoba, Canada, 1984.

136

SUMMARY OF SESSION III: EXPERIMENTS AND SPECIAL EFFECTS

The papers in Session III were all based on experimental studies and generally focused on interpretation of underlying mechanisms of important processes (e.g., radiolysis, dissolution and leachate transport). The Canadian work (Shoesmith) in this area seems particularly significant. It involves the continual development of novel methodology and imaginative planning of long-timescale experiments (lasting several years) that are important in this area of research. The experimental studies themselves can be classified as either simple, in which a single process is examined, or integral, in which more complex and hopefully realistic combinations of processes are studied. From the papers presented, it is evident that these experiment types are complementary and, if coordinated, can yield much more valuable information than either could in isolation.

Two general observations can be noted from the session. First, experi­mental studies of the near field are extremely medium-site and disposal design-specific. Although some experimental techniques may be commonly applicable, important near-field phenomena for, say, disposal of spent fuel in the unsaturated zone in tuff, mixed I/HLW in salt and vitrified HLW in granite are likely to be very different. Second, despite the elaboration of data requirements by modelers in the preceding sessions, very little direct feedback from experimentalists to model development was evident. In order to specify "realistic" experimental conditions and model processes, much more extensive two-way communication is required. These two considerations bring out the need to tailor each experimental effort to fit both the specific environment and system being studies and the models being used to describe them.

137-138

• SYSTEM TESTS FOR STUDYING THE RELEASE OF RADIOACTIVITY FROM HIGH-LEVEL WASTE FORMS UNDER GEOLOGICAL DISPOSAL CONDITIONS

M.N. Gray, R.B. Heimann, J.C. Tait and A.G. Wikjord Waste Management Division

Atomic Energy of Canada Limited Whiteshell Nuclear Research Establishment

Pinawa, Manitoba ROE lLO Canada

ABSTRACT

Canadian studies are focused on a multibarrier concept of disposing of high-level wastes in plutonic rock. There has been considerable effort to specify the components of the underground vault and the physical, geochemical, mechanical and radiological conditions of the disposal environment. The post­closure assessment of the disposal concept requires a reliable prediction of the flux of radionuclides from the vault to the geosphere as a function of time. This paper describes the multicomponent system tests and facilities designed to determine the performance of the engineered barriers of the vault under geological disposal conditions.

1. INTRODUCTION

Canadian research and development on the disposal of high-level nuclear waste is currently in a concept-assessment phase. The disposal concept receiving the greatest attention is based on a system of engineered barriers in a deep underground vault surrounded by a massive natural barrier of. plutonic rock. The engineered barriers include the waste form consisting of either used CANDU fuel or solidified wastes arising from the recycling of used fuel, a corrosion-resistant container, and an impermeable buffer material surrounding the container. The engineered components will be emplaced in boreholes in the floor of the underground vault. The post-closure assessment of the performance of the vault requires a reliable prediction of the flux of radionuclides at the vault-geosphere interface (i.e. a source term) over a period of thousands of years.

In 1960, researchers of Atomic Energy of Canada Limited initiated a pioneering field test to evaluate the performance of a vitrified high-level waste form in a natural sOil-groundwater environment at ambient temperature. Twenty-five, 2-kg hemispheres of radioactive aluminosilicate glass (15 CaO: 9 Na20:4 K20:20 A1203:52 8i02 , wt%) containing 21 TBq (570 Ci) of mixed fissIon products were buried in a sandy-soil aquifer at the Chalk River Nuclear Laboratories. The release and migration of 8r-90 and Cs-137 has been derived from the record of field data and compared to laboratory leaching studies of a hemisphere retrieved in 1978 [1]. After 15 years of burial, the leach rate of the glass decreased from its initial value by more than 3 orders of magnitude. The field data have been interpreted on the basis of a protective layer on the glass surface [2]. A leaching model has been proposed which describes the loss of radionuclides by diffusion from a nondissolving waste form through an inert mineralized layer into an open water system [3].

In the last decade, the concept of a multibarrier vault deep within plutonic rock has evolved, based on extensive laboratory tests on various waste forms, containers and buffer materials. Today", the experimental program is focusing on multicomponent tests under the conditions anticipated in an underground disposal vault. This report describes the laboratory facilities and equipment that have been constructed at the Whiteshell Nuclear Research Establishment (WNRE) to carry out such tests. The facil.iUes include:

1. lead-shielded cells with remote manipulators for the fabrication and testing of high-level waste forms (located in the WNRE hot cell facility);

2. concrete canisters for interactive tests involving radioactive waste forms and other vault components (located in a canister storage area of the WNRE hot cell facility);

3. concrete canisters containing used U02 fuel to study the behaviour of materials in a gamma field (located in the WNRE waste management area);

4. an experimental facility containing used Th02 fuel to study the behaviour of materials in a gamma field (located in the fuel storage pool of the WR-1 reactor, a 40-MWt, organic-cooled, D20-moderated research reactor at WNRE);

140

5.

2.

an underground research laboratory constructed in plutonic rock in the Precambrian Shield (located on the Lac du Bonnet batholith a few kilometers from WNRE).

REFERENCE VAULT CONDITIONS

The physical, chemical, mechanical, radiological and geological conditions expected in an underground vault for high-level wastes will depend upon many factors, including the choice of host rock and engineered barriers; groundwater chemistry; type, quantity and loading of the waste form; and size, depth and configuration of the vault. The anticipated vault conditions are subject to revision as new laboratory and field data are generated and the disposal strategy is optimized. The vault conditions currently assumed in conceptual engineering studies ~re: 2ioreh~le emplacement at 1000 m depth in granitic rock; highly saline Na ,Ca ,CI -type groundwater; 10 MPa hydro­static pressure and 30 MPa lithostatic pressure; high-level waste forms comprising used fuel or vitrified recycle wastes; titanium containers surround­ed by a mixture of 50% sand and 50% clay; maximum container skin temperature of 100oC. In the research programs, the effects of variations in parameters and components are being investigated and reviewed.

3. INTERMEDIATE-LEVEL SHIELDED CELLS FOR FABRICATION AND TESTING OF RADIOACTIVE WASTE FORMS

To understand the possible reactions and interactions involving radio­nuclides, waste forms, container and buffer materials, rock and groundwater, it is necessary to conduct experiments with waste forms loaded with radio­active fission products and actinides. In such experiments, the radionuclides serve as tracers which can be used to help elucidate the processes of dissolu­tion and alteration of the waste form, as well as migration, sorption and precipitation of the released radionuclides.

"-

Irradiated CANDU fuel pellets are readily available in the Canadian Nuclear Power Program and fuel recycle waste forms can be fabricated from limited quantities of high-level liquid wastes arising from small-scale, fuel­recycle experiments. To carry out the fabrication and testing of radioactive waste forms, an experimental facility consisting of five intermediate-level lead-shielded cells has been constructed as an extension of the existing hot cells at WNRE. Each cell (Figure 1) is mounted on a structural steel bench, equipped with remote manipulators, and provided with active drainage, ventila­tion, process chemicals, vacuum, compressed air, and a Halon fire-suppression system. Shielding is provided by 100-mm thick interlocking lead bricks de­signed to allow safe handling of a maximum of 590 GBq (16 Ci) of mixed fission products.

Three of the cells (Figure 2) are interconnected with transfer ports and equipped for the fabrication and testing of high-level waste glasses and glass-ceramics. The required operations include feed preparation, melting, annealing, crystallization and leaching. The fourth cell is equipped to study the dissolution behaviour of irradiated fuel in aqueous systems under both static and dynamic flow conditions at temperatures up to 150oC. The fifth cell is provided with titrimetric, colorimetric, potentiometric, ion-exchange and chromatographic equipment for the chemical analysis of leachate solutions. Samples are transferred out of the facility for mass spectrometry, atomic absorption spectrophotometry and radiochemical analyses.

141

SERVICE PORT

REMOVABLE ROOF PANELS

Figure 1: Schematic of a lead-shielded cell for intermediate-level experiments

Figure 2 : Photograph of three interconnected intermediate- level cells .

142

4. MULTICOMPONENT SYSTEM TESTS USING SHIELDED CONCRETE CANISTERS

High-activity experiments are being conducted in shielded canisters constructed of concrete and steel. These free-standing modules provide a more cost-effective approach for long-term experiments than the use of space within hot cells. Two types of experiments are being carried out. In the first, described elsewhere [4], an irradiated fuel bundle provides a gamma-field to study the corrosion behavior of metallic container materials. The shielded canisters are designed to accept one 22-kg, 5-year cooled CANDU fuel bundle with a burnup of 860 GJ/kg U (10,000 MWd/MgU) [4].

In the second type of experiment, radioactive waste forms (used fuel and glasses or ceramics loaded with fission products) are sealed in pressure vessels along with other engineered and geological components of an under­ground vault. The experiments are statistically designed in order to assess the relative importance of the buffer and container materials, the ground­water, and the ratio of the surface area of the waste form to the volume of the solution (SA/V) on the durability of the waste form. Figure 3 is a cut­away drawing of a canister with a removable secondary containment vessel (SCV) and retaining basket designed to accommodate up to 18 individual pressure vessels made from ASTM grade 2 or grade 12 titanium. Figure 4 shows three pressure vessels mounted in a retaining basket in front of the SCV. A piezoresistive pressure transducer is attached to the head of one of the vessels. The S-shaped service penetrations indicated in Figure 3 accommodate a ventilation line from the SCV and electrical leads connecting thermocouples and transducers to a computerized, data-acquisition system. Each titanium vessel is assembled in a hot cell, prepressurized with argon, and loaded into the basket within the SCV as shown in Figure 4. The SCV is placed in the shielded concrete canister according to the procedures described in [5·].

Figure 5 shows a schematic cross-section of a pressure vessel lined with a cup fabricated from granite. A "minican" made from ASTM grade 2 titanium holds either used fuel or radioactive glass, sandwiched between pellets of compressed clay and metal discs. This waste package is in contact with groundwater of ionic strength simulating either fresh granitic ground­water at shallow depths, or highly saline groundwater of the type found at 500 to 1000 m depth throughout the Canadian Shield [6]. Thus, the waste package configuration resembles a "minivau1t," containing all components expected to be present in an actual disposal vault.

Table I gives the range of the variations of the parameters in the statistically-designed experiments. Nine parameters are considered, the last four of which are held constant within one concrete canister.

At the end of each experiment, the leach solutions will be analyzed for fission products and actinides. The surfaces of the waste form, metal and rock will be investigated using a shielded scanning electron microscope fitted with an energy dispersive X-ray analyzer. Surface scrapings from the waste form will be analyzed by a shielded X-ray diffractometer.

From the analytical results, the leach rates or solubilities of nuclear waste forms will be estimated for conditions resembling those in a disposal vault. Also, the parameters which have a major influence on the leaching and dissolution behaviour of high-level waste forms will be identified.

143

Figure 3:

Pressure Vessel

Cut-away of a concrete canister for high-level experiments.

.-~.

- 2.08

Secondlry Contlinm.nt V .... I

R.tlininll Belk.t

-::: Servic. P.n.trllionl

H •• t.r

VENT VALVE - CHARGING VALVE ;~-£,~::U_WITH DIPTUBE

AND PERCOLATOR

COPPER GASKET~H~~~~~-tl_ROCK LID

TITANIUM, INCONEL

GROUNDWATER ~~~~~-('BUFFER,SANDI

1I4----lIlI-i-b,..+- PERFORATED MINICAN

(CANISTER OR OVERPACK 1~~*'Iif::=~1 1 1P-~4~ COMPRESSED BUFFER

WASTE GLASS""-{"""~F4:

Figure 5:

C;=3~~-ROCK CUP

Pressure vessel with "minican" to simulate the components of a disposal vault.

144

Figure 4: Pressure vessels mounted in a ret ain ing basket 1n front of a secondary containment vesse l .

Table I: Experimental Parameters Varied During the Research Project

Parameter

(i) Waste form

(U) Buffer material [7,8]

(iii) Groundwater [9]

(iv) Container material [10]

(v) SA/V ratio

(vi) Rock material

(vii) Pressure

(viii) Reaction time} (ix) Temperature

Range of Variation

Used UOZ fuel Fuel recycle waste glass

Ca-montmori11onite (CECa = 1360 meq/kg) Illite (CEC = 160 meq/kg) Na-montmori1lonite (CEC = 8Z0 meq/kg)

Granite groundwater (Ib = 1.4 x Standard Canadian Shield Saline (SCSSS) (I = 1. 37 mo l/L)

ASTM grade-Z titanium ASTM grade-1Z titanium Incone1 6Z5

1Z - 1Z0 m-1

Granite Gabbro

-3 10 mo1/L) Solution

10 MPa (hydrostatic, 1000-m depth) 30 MPa (lithostatic, 1000-m depth)

6 months at ZOOoC 6 months at 1500 C

1Z months at 1000 C Z4 months at 500 C

a) CEC = cation exchange capacity b) I = ionic strength

5. EXPERIMENTS TO INVESTIGATE EFFECTS OF GAMMA RADIOLYSIS

Several studies are underway to assess the relative importance of solution radio1ysis on radionuc1ide release from candidate waste forms. Since groundwater close to a waste package will ini£~1lly be subjected to gamma radiation from fission products (principally Cs), it is necessary to under-stand the radiolysis processes which could enhance container corrosion and radionuclide release from the waste form by such mechanisms as changes in radionuclide valency or production of reactive radiolysis products.

Several waste forms, doped with U and inactive Cs, Sr and La, were sealed in silica capsules containing the various components of a mu1tibarrier disposal environment (Table II). The capsules were put into titanium pressure vessels, placed in a gamma irradiation facility consisting of a shielded concrete canister with an inner basket holding up to 1Z experiments (Figure 6)

o and heated to 100 C for 485 days. Used UOZ fuel elements provide a gamma dose rate of 400 R/h inside the titanium pressure vessels. The 1eachants in the capsules were analyzed for the major constituents and dopants, and surface

145

analyses were performed on the specimen coupons us'ing secondary-ion mass­spectrometry and scanning electron microscopy. Initial results are given in [11] •

A second facility (Figure 7) for studying radiolysis effects is nearing completion. The facility will be located under 5 m water in the fuel storage bay of the WR-1 experimental reactor and will use a hollow-core, fa~t­negtron, Th02-fuel assembly to provide a gamma dose rate in the range of 10 to 10 R/h. A pressure vessel containing sealed experimental cagsules will be filled with a thermal transfer fluid and heated to 100 to 150 C. Sealed capsules containing components shown in Table II will be removed periodically so that radiolysis effects can be followed as a function of time. The experiments will investigate the fundamental aspects of radiolysis in simple systems and provide a basis for modelling radiolysis effects under vault conditions.

Table II: Components Used in Gamma-Radiolysis Experiments

Component

Waste form

Leachan t (%, 7 mL)

Container

Buffer

Rock

Atmosphere ( 'V 0.1 kPa)

Type

Glass - aluminosilicate, borosilicate Glass ceramic - titanosilicate

Deionized water Granite groundwater Standard Canadian Shield Saline Solution

Stainless steel Titanium

Sodium bentonite clay

Granite Gabbro

Air Argon

6. IN-SITU EXPERIMENTS PROPOSED FOR AN UNDERGROUND RESEARCH LABORATORY

An underground research laboratory (URL) is being constructed in a previously undisturbed batholith in the Canadian Shield near WNRE [12]. Multicomponent experiments on the interactions between nonradioactive waste forms, engineered barriers, granitic rocks and groundwaters will be conducted in the URL. These multicomponent experiments are currently planned to begin after August 1986 when URL construction will be complete.

Metallic containers will be placed in vertical boreholes in the vault floor and separated from the host rock by a compacted clay-based buffer [13]. The buffer material will support the container and minimize the rate at which radionuclides migrate from a breached container to the geosphere. It has been suggested [14] that the principal component of the buffer material should be

146

T

Sand

Silica

E I"t)

an

capsule --t---+~.~I

2.6m ---~------.I

"I Concrete

r- ~

Insulation

Heater

Steel basket

Pressure vessel

used fuel bundle

Figure 6: Shielded concrete canister irradiated internally by a U02 fuel bundle

5m water pool

IOOcm

Used Th02 fuel

Fuel string support tube

Contam i nation shield

Pressure Vessel

I~I-II--+- Heating Rod

*+~J.lI-+-Silica Sample vessel

~I+l-II--l-- Organic fluid reactor coolant

Figure 7: Gamma irradiation facility located in a fuel storage pool for the WR-l reactor.

147

4-5m

Materials handling --+----+1'

system

cell cop --+-----..

Bitumastic seal

Backfill material

rock mi~ture) ( cl ay, crushed ~+t..,.f-~-1+Hrl~~1"""

Rock anchors ~~"""'I

Heater / container

Sand infill

Forced ventilation system and lor services

Load cell (donut type) on rock anchors

Boreholes for piezometers and thermocouples

Bentonite / sand buffer material

Figure 8: Design for buffer mass tests in boreholes of the underground research laboratory.

148

sodium bentonite because it retards the migration of radionuclides and swells on exposure to water. Swelling will lead to self-healing if fractures develop in the buffer. Available data [15,16] indicate that the important physical properties of the buffer are optimum when the bentonite clay is mixed with a graded sand in a mass ratio of approximately 1:1.

Tests to determine the nature of the physical interaction between the container, buffer, host rock and groundwater are now being designed for empla­cement in the URL. A conceptual design for the first experiment is shown in Figure 8. The granitic rock, the buffer and the container will be extensively instrumented to provide a detailed history of changes in the total stresses, pore water pressures, temperatures and moisture distributions within, and around, the buffer. In addition to the measurements required to describe buffer behaviour, consideration is being given to the inclusion of simulated waste glass coupons within the buffer mass. The coupons will be examined be­fore and after the test by surface-analysis techniques to determine the extent of interaction between the glass and the other components of the experiment.

Alternative test configurations for including waste forms within the buffer material can be considered, provided they do not interfere with the buffer mass tests. For example, it may be possible to include small heated "pineapple" test assemblies, in which glass coupons are placed in contact with various buffer and/or container materials, either within the buffer mass or in a small diameter hole drilled at the base of the emplacement borehole.

7. SUMMARY

Canada has carried out the world's longest field evaluation of a dur­able active waste glass. Interactive tests involving waste forms, engineered barriers, groundwaters and rocks are being carried out under simulated disposal conditions in shielded concrete canisters. In-situ, nonradioactive tests in an underground research laboratory are being planned. The tests will improve our understanding of the behaviour of the man-made materials in an underground disposal vault and help determine source terms for the environmental and safety assessment of the vault.

REFERENCES

1. Merritt, W.F.: "The Leaching of Radioactivity from Highly Radioactive Glass Blocks Buried Below the Water Table: Fifteen Years of Results", Proceedings of a Symposium on the Management of Radioactive Wastes from the Nuclear Fuel Cycle, International Atomic Energy Agency, ~, 27, 1976.

2. Melnyk, T.W., Walton, F.B., and Johnson, L.H.: "High-Level Waste Glass Field Burial Test: Leaching and Migration of Fission Products", Nuclear and Chemical Waste Management, ~, 49-61 (1984).

3. Harvey, K.B. and Litke, C.D.: "Model for Leaching Behavior of Alumino­silicate Glasses Developed as Matrices for Immobilizing High-Level Wastes", J. Am. Ceram. Soc., ~(8), 553-556 (1984).

4. Crosthwaite, J.L.: "The Immobilized Fuel Test Facility Program", Proceed­ings of the 16th Information Meeting of the Nuclear Fuel Waste Management Program, Atomic Energy of Canada Limited Technical Record, TR-218*, 1984.

149

5. Heimann, R.B. and Johnson, L.H.: "Design of Multicomponent Systems Tests on High-Level Nuclear Waste Forms", Advances in Ceramics, Vol. 8 (Nuclear Waste Management), G.G. Wicks and W.A. Ross (ed.), 337-345, 1984.

6. Fritz, P. and Frape, S.K.: "Saline Groundwaters in the Canadian Shield - A First Overview", Chem. Geol. ~, 179 (1982).

7. Wardrop, W.L. and Associates Limited, in Association with Canadian Mine Services Limited, and Hardy Associates (1978) Limited, "Buffer and Back­filling Systems for a Nuclear Fuel Waste Disposal Vault", Atomic Energy of Canada Limited Technical Record, TR-341*, 1985.

8. Quigley, R.M.: "Quantitative Mineralogy and Preliminary Pore-Water Chemistry of Candidate Buffer and Backfill Materials for a Nuclear Fuel \~aste Disposal Vault", Atomic Energy of Canada Limited Report, AECL-7827, 1984.

9. Sargent, F.P.: "Applied Chemistry and Geochemistry Research for the Canadian Nuclear Fuel Waste Management Program", Atomic Energy of Canada Limited Technical Record, TR-149*, 1982.

10. Nuttall, K. and Urbanic, V.F.: "An Assessment of Materials for Nuclear Fuel Immobilization Containers", Atomic Energy of Canada Limited Report, AECL-6440, 1981.

11. Tait, J.C.: "Gamma-Radiolysis Effects on Leaching of Nuclear Waste Forms: Influence of Groundwaters and Granite on Gaseous Radiolysis Products", submitted to Radiation Physics and Chemistry.

12. Simmons, G.R., Brown, A., Davison, C.C. and Rigby, G.L.: "The Canadian Underground Research Laboratory," Proceedings of the lAEA International Conference on Radioactive Waste Management, Seattle, 1983.

13. Bird, G.W. and Cameron, D.J.: "Vault Sealing Research for the Canadian Nuclear Fuel Waste Management Program," Atomic Energy of Canada Limited Technical Record, TR-145*, 1982.

14. Oscarson, D.W. and Cheung, S.C.H.: "Evaluation of Phyllo-Silicates as a Buffer Component in the Disposal of Nuclear Fuel Waste," Atomic Energy of Canada Limited Report, AECL-7812, 1983.

15. Gray, M.N., Cheung, S.C.H. and Dixon, D.A.: "Swelling Pressures of Compacted Bentonite/Sand Mixtures," Proceedings of Materials Research Society Symposium on the Scientific Basis for Nuclear Waste Management, Boston, 1984.

16. Dixon, D.A., Gray, M.N. and Thomas, A.W.: "A Study of the Compaction Properties of Potential Clay-Sand Mixtures for Use in Nuclear Fuel Waste Disposal," Proceedings Int. Symp. on Clay Barriers for Isolating Toxic Chemical Wastes, Elsevier Scientific Publishing Co., Amsterdam, 1984.

* Unrestricted unpublished report, available from SDDO, Atomic Energy of Canada Limited Research Company, Chalk River, Ontario, KOJ 1JO

150

STUDY OF THE RELEASE OF VARIOUS E~NTS FROM HLW GLASSES IN CONTACT WITH POROUS MEDIA

F. Lanza, E. Parnisari, C. Ronsecco Commission of the European Communities

Joint Research Centre - Ispra Establishment Materials Science Division

21020 Ispra (Va) - Italy

ABSTRACT

When a borosilicate glass incorporating HLW is embedded in a static po­rous media the leached elements diffuse into the media. The release of the various elements from the glass is governed by different mechanisms. Soluble elements (such as Cs or Tc) will be released as a direct consequence of the glass degradation. The release of elements which are less soluble than silica, on the contrary, will be dominated by the solubility limit and by the subse­quent diffusion in the porous media. An integral experiment has been devised which allows at the same time to measure the leaching rate and the diffusion of the various species in the porous media.

151

INTRODUCTION

In the description of the release of radionuclides from the conditioned waste it is customary to speak of a near field region. In this region are coupled the phenomena of corrosion, leaching, complexation reaction, diffusion and reprecipitation. It is difficult to define physically the boundaries of the near field. Certainly they comprehend the engineered part of the reposi­tory.

Many concepts of final repository are actually at the study. In most of the concepts in between the glass container and the repository wall, a com­pacted porous material is foreseen~ The use of a backfilling porous material is generally accepted for both granite and clay repositories. In the case of clay repository or of a subseabed emplacement even the main barrier is a wet porous material. It appears then that it is worthwhile to study the mecha­nisms which dominate the leaching of the glass and the release of the various elements.

As far as leaching of the glass is concerned a porous material filled by interstitial water presents aspects which are quite peculiar. The system cannot be considered equivalent to a static condition due to the fact that by diffusion the product released by the glass can migrate. Possible saturation effects depend then on diffusion rate. Nor can it be considered equivalent to a dynamic condition as the migration rate, depending on adsorption and diffusion, varies with the various elements. It appears then that a leaching test specific for glass embedded in a porous media must allow not only to evaluate the glass degradation but also the release and diffusion of the va­rious elements.

PRELIMINARY EXPERIMENTS

In a first study aimed at the understanding of the main mechanism of leaching in porous media, a very simple leaching system was utilized. The glass investigated was a borosilicate glass containing 20% in weight of simu­lated waste oxides. Its silica content is 4S%; the detailed composition is reported in [1].

Different types of wet porous media were utilized. Firstly a series of tests were conducted with a mixture of montmorillonite and sand. Other tests were conducted using a mixture made of montmorillonite, sand and Fe203. From these mixtures a paste was obtained using distilled water. A second series was conducted using only sand wetted as before with distilled water. Finally a series of tests has been conducted using red clay sea sediments extracted from the CV2 site wetted by sea water.

The experiments were conducted in glass ampoules in which 30 cc of a di­lute paste was introduced. Over the paste an equal volume of water was placed and then the ampoules were sealed. Experiments were performed at 30, 50 and SOoC. Every three weeks an ampoule was removed and the sample extracted. The tests were terminated after 60 weeks. More details on those tests are given in [1].

The weight loss data obtained at SO and 300 C using distilled water are

152

reported in Figs. 1 and 2. The data are exposed in a log-log diagram. A li­near relationship is the expression of a relation of the type ~W/S = atn. As Lutze [2] pointed out a value n = 1 is representative of a dissolution attack while a value n = 0.5 underlines the predominance of diffusion mecha­nisms. In the three diagrams are also reported data corresponding to weight losses in slowly renewed water 'which were obtained previously [3]. We will consider these data as a basis for evaluation if saturation effects are present.

At SOoC it can be seen that the attack of the glass in montmorillonite and montmorillonite + Fe203 in a first time follows closely that occurring in slowly renewed water. Both seem to follow a dissolution kinetic. After about 100 days the leaching rate changes passing to a kinetic depending on saturation controlled by diffusion phenomena. The glass sample immersed in sand not only shows a lower leaching rate, but also from the beginning, a ki­netic dominated by saturation and diffusion phenomena. The relatively high scattering of the data has to be attributed mainly to the very low weight losses. Lowering the temperature the attack in montmorillonite tends to in­crease in respect to slowly renewed water. We note that the weight losses in montmorillonite are practically the same at 50 and 80oC. In the presence of montmorillonite dissolution seems the predominant mechanism. Tests in sand were conducted at 500 C and showed the same picture as in the tests at 80oC, low attack rate and diffusion predominance.

A first interpretation of this complex behaviour can be attempted con­sidering as a main phenomenon of leaching the dissolution and absorption of Si02. In a porous system, we can assume that the interstitial water in con­tact with the glass reaches rapidly the saturation limit. As a consequence, the silica dissolution rate will be governed by the diffusion of silica in the porous media. For the case of montmorillonite it has to be noted that clay can absorb silica. Following the data of Tau [4] at room temperature the ini­tial Kd value is around 200 cm3/g. As a consequence the silica released by the glass is readily adsorbed by the montmorillonite and the glass behaves like in renewed water following a direct dissolution kinetic.

Only after a certain time, when sufficient silica has been released to saturate the montmorillonite surrounding the glass, saturation appears and diffusion of silica begins to be the rate determining step. Such an effect is easily apparent in the data at SOoC. In montmorillonite at lower temperature the weight losses are even higher than in the case of the slowly renewed water, indicating that probably silica adsorption interferes in the process of the surface gel formation.

The importance of the dissolved silica in the leaching phenomena is un­derlined by the tests performed with the glass immersed in sand. In this case the interstitial water contains yet an elevated amount of dissolved silica; as a consequence the dissolution potential is lowered and weight losses are very low.

The tests performed with Fe203 do not differ appreciably from those with montmorillonite. McVay [5] has noted that the presence of iron tends to accelerate the leaching process due to the formation of a highly insoluble silicate. In our case it seems that such an effect is covered by the influence of absorption of the clay. Tests with real corrosion products are needed in

153

order to verify this point. In Fig. 3 the data referred to leaching in red clay sea sediments are shown. Sea sediments present an interesting phenomenon. At their original temperature (2-40 C) they are in equilibrium with the inter­stitial water which has an appreciable concentration of dissolved silica, ~ 10 ppm. When tests are performed at higher temperature the system sea water­sediment is temporarily disturbed. At 30 and 500 C the disturbance is not so strong to avoid saturation effects and a diffusion system seems prevalent, but at 800 C, where the unbalance is maximum, a preliminary period of dissolu­tion is noted.

In order to verify if the increase in temperature influences also the tests performed with a mixture of montmorillonite and sand, a test was de­vised in which the wet mixture was pre-treated for six months at 800 c. After­wards a test was performed at 500 C as usual. The results are exposed in Fig. 4. It can be seen that the pre-treatment of the mixture changes the kinetic of the weight losses, probably due to saturation effects. It appears clearly that in performing the experiments, care should be taken in trying to be in a condition of equilibrium of the system porous media-interstitial water at the temperature of the test.

COLUMN EXPERIMENTS

Preliminary experiments have shown the importance of diffusion phenomena in the evaluation of the rate of degradation of the glass. Besides, it ap­peared clearly that it was necessary to be able to evaluate the release of the various elements which are not released congruently. It seemed necessary, then, that the leaching apparatus could allow to measure, at the same time, the diffusion of the released elements.

In the diffusion experiments it is essential to have a well defined geo­metry of the system in order to be able to apply simple mathematical models. In general a cylinder of the porous matter pressed against a source is used. In such experiments the source term is a sheet of filter paper soaked in a solution containing a radioactive element [6]. Following these lines a design has been developed. In this apparatus a cylinder of porous material wet by its interstitial water is pressed by a piston against a glass block [7]. In order to avoid that the porous material dries during the experiments, it is necessary to keep it in a tight system. For this reason the piston is equipped with two O-rings while in the lower part the porous material is kept tight by the use of an O-ring which is pressed against the shoulders of the crucible. The system is tested at the helium leak detector and i.t is considered accep­table if it shows a leak lower than 10-8 lis. In order to keep constant in time the contact pressure between the wet porous material and the glass sample, the upper piston is pressurized by a hydraulic system up to a maximum pres­sure of 2.5 MPa. Every sample is extracted successively every three weeks. The extracted unit is then dismounted in order to recuperate the glass sample.

When the glass in put in contact with the clay the surface layer which results from the leaching process can easily absorb products contained in the clay. As a consequence the weight loss measurements are not very significant. It was decided to evaluate the weight losses after having taken out the gel layer and consider this value as an index of the glass degradation rate. The

154

cylinder containing the porous material after the test is positioned in an apparatus which allows, using a micrometric screw, to extrude progressively the porous material from the cylindrical body. The extruded part can be sliced, using a blade, and analysed for the diffused elements. This system allows to separate slices of a minimum thicknesss of 0.5 mm. The slices of clay are analysed by neutron activation techniques.

A preliminary series of tests has been performed using Boom clay at the temperature of 750 C up to a maximum time of 106 days. The glass degradation follows a linear dependence from time. Even taking into account that the sur­face layer has been removed, the values are rather high and higher by a fac­tor of 5 than those obtained in renewed water. A linear regression gives as a best estimate the following relationship:

~W/S = 8.6.10-5 t (2)

If we compare these results with those obtained previously in the ampoule tests it appears that the times of exposure used allow to obtain data referred to periods in which the absorption of the released silica is prevalent. An ex­perimental evidence of the influence of diffusion on the rate of degradation probably needs a longer leaching time.

In order to be able to evaluate the release and diffusion of the various elements it is worthwhile to classify them following their specific leaching rate and their solubility limits. If we assume that the degradation rate is dependent from the leaching rate of silicon, which composes the basic struc­ture of the glass, we can define the following three classes: i) elements having a leaching coefficient higher than that of silicon; ii) elements having a leaching coefficient equal to or lower than that of

silicon and a global solubility limit higher than that of silicon oxide; iii) elements having a leaching coefficient equal to or lower than that of

silicon and a global solubility limit lower than that of silicon oxide.

In the first class are comprised the alkalies, which can be released from the glass by interdiffusion with H30+. Cesium is the most important ele­ment in this class. However, in normal conditions for complex glasses the re­lease rate due to diffusion is not prevalent in respect to the degradation rate. Only for conditions of advanced saturation alkalies can be released at a higher rate than silicon. We will consider, then, the first class together wi th the second one.

Elements of the second class, having an elevated solubility, as soon as they are freed from the glass, due to the breakage of the polimeric Si-O-Si network, begin to diffuse. Their release rate will be proportional to the degradation rate of the glass. We have seen previously that in this series of tests the degradation rate is constant during the time of exposure. As a con­sequence the elements of the second class will be released as a constant flow during the exposure time.

Elements of the third class even if freed by the glass degradation once reached the solubility limits of the various possible chemical species will not be released anymore. The release of the third class, then, will be domi­natedby the diffusion of the various possible chemical species and will be independent from the glass degradation rate.

155

The release of the elements of the first and second class corresponds to the diffusion equation

dC d 2C at = D dX2

having as a boundary condition

dcl - D - = ~ dX X=O

(3)

(4)

where ~ is the constant flux at the surface. Assuming that the column is in­finite and that thickness of the leached glass is negligible in comparison with the diffusion path, using the method of the Laplace transform, it is pos­sible to obtain the following concentration distribution

c ::;: ~/D[hDthr • exp(-x2/4Dt) - x erfc (x/2Dt)] (5)

The concentration at the interface will be

(6)

The distribution of iron (Fig. 5) into the clay follows closely the distribu­tion described by eq. (5). For elements of the third class the diffusion equa­tion will have the boundary condition

cl = c x=O 0

The distribution of the concentrations of the element is given for this case [8] by the following expression

c = c erfc (x/2mt) o

The amount of the element released by the glass is

M = 2c IDt/7r o

(7)

(8)

( 9)

It has to be underlined that for this class the amount released is indepen­dent from the rate of degradation of the glass. The release is governed by the solubility limit of the migrating species and by its diffusion coefficient. Uranium distribution into the clay (Fig. 6) follows closely the distribution described by eq.(8). Cesium shows a high diffusion coefficient in clay so that for the tests made at longer times it was not possible to accept the ap­proximation of an infinite column. The solution proposed by Crank [8] for a sheet having a continuous diffusion flux F at the surface has been utilized. Using a best fitting routine the concentration data for Cs, U, Co, Fe, La, Ce, Eu and Hf have been analysed in order to obtain the corresponding diffu­sion coefficient. Table I gives the obtained results.

156

TABLE I - Diffusion coefficients in clay

Element D (cm2/s) Element D (cm2/s)

Cs 5~2·1O-7 La 1.8.10-9

U 1.2.10-8 Ce 1. 10-9

Co 4.8.10-9 Eu 3.9,10-10

Fe 2.10-9 HF 7.6.10- 11

CONCLUSIONS

The column experiments allow to study at the same time the leaching of the glass and the diffusional release of the various elements in the sur­rounding porous media. It appears clearly from the experiment that in consi­dering the glass as a source of the various elements, it is necessary to make a distinction between soluble species, which are released following the kine­tic of the glass degradation, and insoluble ones which are released by solu­bilization and diffusion. For these elements, then, the evaluation of their release is strictly linked to the studies on speciation and to the evaluation of the solubility limits.

A continuation of the study is needed in order to evaluate the influence on the degradation rate of a pre-treatment time at test temperature. The influence of corrosion products on the degradation rate and on the diffusion of the various elements is also to be investigated.

ACKNOWLEDGEMENTS

The authors wish to acknowledge the courtesy of Mr. Bonne (SCK/CEN) for the supply of the Boom clay, the assistance of Mr. Pietra for the nuclear activation analysis, and the assistance of Mr. Saltelli for the analytical treatment of the diffusion equations.

REFERENCES

1. Lanza, F. and Ronsecco, C.: "Glass leaching in wet porous media", EUR 9215EN (1984).

2. Altenheim, F.K. and Lutre, W.: "The mechanism for hydrothermal leaching of glass and glass ceramic nuclear waste forms - Scientific basis for nuclear waste management", Vol. 5, North Holland (1981).

3. Lanza, F. and Parnisari, E.: "Influence on film formation and its com­position on the leaching of borosilicate glasses", Nuclear and Chern. waste Management, Vol.2, pp.131-137 (1981).

4. Tau, K.H.: "The effect of interaction and adsorption of silica on struc­tural changes in clay minerals", Soil Science 134, Nov. 1982, 300.

5. MCVay, G.L.: in Proc. MCC Workshop on Leaching Mechanisms of Nuclear waste Forms, ed. J.F. Hendel, PNL-4382 (Aug. 1982).

6. Schreiner, F., Fried, S. and Friedman, A.M.: "Measurement of radionuclide diffusion in ocean floor sediment and clay", Nucl. Tech. 59, p.429,

157

Dec. 1982. 7. Lanza, F., Demicheli, U. and Parnisari ,~.: "Development of a test

method for leaching glasses i~ersed in porous media", EUR 8921EN (1984). 8. Crank, J.: "The mathematics of diffusion", p.61, Clarendon Press (1983).

158

5

0,5

o

0,1+~~----------, ........ ------------l 10 100

t Idays)

1000

Fig. 1 Specific weight losses at 80 0 e in various porous media: • renewed water; A mont + sand; 0 sand.

6

0,1

o

, 50

, 100

t Idays)

I

500 1000

Fig. 2 Specific weight losses at 30 0 e in various porous media: • mont + sand; 0 renewed water.

159

5

./

• /'

.. / 7·

y o

O,'+--------r----r-------.--~ 1 50 100 500 1000

t (days)

Fig. 3 Specific weight losses in sea sediments at various temperatures: • 80°C; 0 50°C; 630°C .

100 t (days)

• • •

1000

Fig. 4 Leaching at 50°C in montmorillonite sand mixtures: o pretreated at 80°C; • fresh mixture.

160

10000

o~ E

\\ Q. • Q

\ -;; 1000

• 0

"" 0

100 5

x (mm)

Fig. 5 Diffusion of iron in clay at ,SOC after 0 63 days; • 106 days.

E ~ u

100

10

1+--~ o

--~-~-~---,-~---r-~--~'---,-5 10

x (mm)

Fig. 6 Diffusion of uranium in clay at 7SoC after 0 43 days; • 63 days; , 84 days.

161-162

LABORATORY SIMULATION OF RADIONUCLIDE GEOLOGICAL MIGRATION

A. Avogadro, G. Bidoglio, A. Saltelli Commission of the European Communities

Joint Research Centre 21020 Ispra (Va) Italy

ABSTRACT

The quantitative description of radionuclide migration in the geosphere demands that the physico-chemical characteristics of species released from the source term are considered. Laboratory experiments which allowed the identification of the migration mechanisms involved aI.'e described; the physical model adopted corresponds to

. porous media permeated with fresh and/or saline ground water. The waste form considered is the vitrified high level waste and problems connected with the trasport of colloidal as well as of soluble species are discussed following a theoretical and experimental approach.

163

1. INTRODUCTION

The dispersion behaviour of radionuclides in the geosphere depends upon their physico-chemical characteristics and two main classes may be identified: colloidal and soluble species. For the soluble species, the interactions will be dependent upon the valence state and speciation of the nuclide while for the colloidal species the size distribution and the electric charge carried by the particles governs their retention.

The programme oeing undertaken by the Joint Research Centre at Ispra in the field of geochemical interactions concerns the laboratory simulation of the release and transport through the geosphere of radionuclides leached from vitrified HLW.

Conditions existing around the Boom clay formation .in Belgium and the Gorleben salt dome in Germany have been chosen as reference cases. In both cases migration towards the biosphere, following release from the repository, may occur through porous argillaceous formations surrounding the disposal site. It is this porous material that is being used in laboratory percolation experiments.

The working methodology adopted is based on a type of feed-back loop established between simulation experiments, physico-chemical forms determina­tion and model development.

In laboratory simulation experiments, the leachant solution flows over a borosilicate glass containing the nuclide to be studied and then through a c.olumn containing the geologic material. At intervals the column is disconnected and the activity profile recorded by gamma-scanning. Percolation experiments under normal atmospheric conditions (oxic conditions) or in special boxes under nitrogen containing 0.1% CO

2 (anoxic conditions) are

performed in order to control the effect of the redox potential of the system.

2. Results and discussions

2.1. Salt rock disposal option

In the salt system condition, the leaching solution (a saturated brine of composition reported in TAB.I), is in equilibrium with the porous geologic material. Figure 1 reports the contamination profiles of Americium (oxic conditions), for three different loading volumes, in a column filled with a typical soil overlying the salt rock dome (95% silica, 5% consisting of clay material, magnetite and rutile) . The observed trend suggests the existence of a saturation effect. Figure 2 shows the breakthrough curve which confirms the saturation of the retention capacity after 250-300 days. The retardation factor R

f determined from the ratio between the velocities of

the water and of the radionuclide or from the ratio of the total activity retained in the column to the activity at the outlet at saturation, ranges around _rOO. The corresponding Kd term consequently assumes a value of 53 ml.g .

164

The ~ttch I(d measured at the same Am concentration of the column influent (75 ml g ) is in good agreement 'vi th the partition coefficient determined from the column parameters. This fact sugges'ts that sorption reactions are at equilibrium in the colunm; the slightly higher Kd in the static system may be a consequence of the break down of soil particles during the shaking procedure (1).

The same experiments performed 'vi th Neptunium doped glass, under oxic condi tions, gave similar results showing that sorption phenomena are the result of reversible equilibrium reactions in this case also. A linear sorpt,ion isotherm, corresponding to a Kd of 8, was in fact determined from ,vhich the same retardation factor of 29 obtained by column data was derived. This indicates an equilibrium reaction but a weaker retention in respect to Americium. In percolation experiments with Neptunium under anoxic conditions, the activity retained by the column increases by a factor 4 corresponding to a H

f of 119. Figure 3 reports schematically the breakthrough curves for the

various cases studied. The results of batch experiments reported in figure 4 were compared with

the curve shape given by conventional sorption models.

The least square analysis of the Langmuir equation eq

1 1 Gm - =. -- +-Kd bK b

gave the best fit allo'i\Dg the eV':1uation of the apparent binding energy K which results in 8,1.10 1. mole for Am and several orders of magnitude lower for Np. The adsorption maxima b of the soil material were also evaluated and compared with the column data. For the Am leachate concentration of

-10 -l) 10 H, t~r value obtained has the same order of magnitude {3.6-7.6.10 m.mole. g ) in both cases indicating the total satur,:yon of the available sites, while for the Np leachate concentration of 1.5.10 M column saturation is reached below the adsorption maximum. Definite quantitative data for Neptunium are not yet available.

The differences observed in column behaviour result from the different characteristics of the interacting species. Owing to the difficulties assiociated with a direct identification of soluble chemical forms of Americium and Neptunium, the equilibrium distribution of species was determined from Ii tera tU1'C data, taking into account only homogeneous chemical reactions. Table II gives a summary of the stability constant values at 1M ionic strength for Americium and Neptunium interaction with the major inorganic ligands in brine. The available data refer to ionic media different than those of interest in this study. The assumption that the activity coefficients do not vary with ionic strength above 1H is an approximation, due to the lack of any reliable model in predicting such a dependence.

The computer modelling output suggests that Am( III) should exist mainly as a mixture of the first two chloro-complexes in brine. Figure 5 shows how

l+ -the percentange of americium chloro-complexes (AmCI +AmCI;) in a 5.4H CI

165

ions solution, varies with pH and ~CO. The complementary vQlume represents the sum of americium carbonate complexe~.

The partitioning of the Np(V) sC)luble species is reported in figure 6 from which it appears that, apart from the Neptunyl ion, the only complexed form present in saturated brine solution is the neutral NpO Cl species. The stronger interaction of Americium compared with Neptuniqm wfth active sites of the soil is clearly the re~;ult of the higher ionic potential of Am soluble species.

2.2. Clay disposal option

In the case of the clay disposal system, a flow of water having composition reported in Table I leaches the simulated vitrified waste and then percolates through the columns filled with the porous strata overlying the Boom clay.

The flow corresponds to a linear velocity of about 27 m per year. The sand layer has 34% porosity, a permeability of about 10-4 mls and contains 20.8 % of glauconite clay mineral.

In parallel with column percolation experiments, ultrafiltration tests of the glass leachate through membranes having different porosities \vere carried out. For the case of Americium the results of filtration experiments are shown in figure 7 while in figure 8 the column profiles in mplar units are reported for three different loading volumes. The Am 510rtent in the water at the column outlet reached the value of (3.5 - 6.6) .10 H very rapidly at the beginning of the loading phase and remained constant throu~Y8ut the experiment while the leachate inlet concentration was fixed at 3.8.10 H.

The large percentage of Am retained on the filtering membranes suggests that the transport is largely dominated by the colloidal species; it is also evident from the smooth shape of the curve in figure 7, that the colloid size is better described as a distribution rather than as a single value. If the trend of the size distribution is assumed to be representative also of the distribution of the filter coefficient in the classical filtration equation (2), a satisfactory fit is obtained between theoretical and experimental data. This is reported in figure 9 for the 170 days percolation experiment. This confirms that for a convective fresh water system the predominating process of retention iIi a porous medium is colloidal filtration.

For percolati~~O experiments shorter than 170 days, a plateau at Am concentration 2.10 1v1 appears in the final portion of the column. This plateau shows two apparently contrasting features a) it is established very rapidly (within 28 days), and b) it does not increase with time. The existence of the plateau cannot be due to the aCJivity in the interstitial water because a concentration factor of about 10 between the plateau and the outlet concentration exists. If this retention were due to colloid filtration 've should observe an increase of the plateau with time, which is not the case. It must also be considered that the Am,soluble plus particulate,concentration in the interstitial water i~lo<?t constant within the colun.:.rb varying between the inlet value of 3.8.10 H and the outlet of 5.10 H, mainly as a

166

consequence of the filtration mechanism. The Am interstitial concentration tends to increase \vith time as the pulse moves within the column.

As the plateau is much likely given by soluble adsorbed species rather than filtered microcolloids, the constancy of the Am concentration on the plateau suggests that the adsorbed Am has reached its saturation value, and it is not affected by any further increase of the Am concentration in the liquid phase. This phenomenon can be described by a Langmuir isotherm. Wi th this assu~f~ion it is possible to verify (2) that a column oulet concentration of 5.10 M §an be reached very rapidly even if the concentration factor is close to 10 .

The column migration pattern of Americium in its various physico-chemical forms present in the leachate is therefore governed by the fixation of the colloidal fraction in the first layers of the column. The behaviour in the remaining portion of the column is the result of the migration of soluble species and their sorption capabilities. A distribution diagramme of Am soluble species as a function of pH for a chemical environment typical of the clay aquifer has been developed (3) and is reported in figure 10.

Under "field conditions" the colloidal fraction will be retained after some meters (or tens of metres), only soluble forms will migrate in a region far from the source and the various soluble ions will be in equilibrium among themselves. When AmCO+ and Am (C0

3)- migrate in equilibrium, the overall Am­

pulse should travel wiih an average r~tention factor governed by the ratio of the equilibrium concentration of the two complexed species present (4).

3. Conclusions

\vith column simulation experiments three predominating processes have been identif:ied:

colloid filtration governed by a distribution of values of filter coefficient; this process is particularly effective during the percolation of a fresh water l~achate through a porous formation a sorption phenomenon adequately describ~d by a Langmuir isotherm; the Americium species reach the maximum adsorption value in both systems analyzed while for Neptunium a linear sorption isotherm was determined for the salt aquifer

escape of unretained species) experimental evidence indicates that in the case of fresh water aquifers, a small fraction of Americium in the form of carbonate complexes flows through the column with very little retention. In the case of a brine aquifer the chlorocomplexes interact with geologic material through equilibrium reactions resulting in retardation factors 0f 204 and 29 respectively for Americium and Neptunium.

167

4. References

1. G. Bidoglio, A. Avogadro, A. De Plano, G.P. Lazzari; Geochemical pattern of Americium in the saline area surrounding a rock salt formation : Radioactive Waste Hanagement and the Nuclear Fuel Cycle, in press

2. A. Saltelli, A. Avogadro, G. Bidoglio; Americium filtration in glauconitic sand columns Technology, Nuclear Technology Vol. 67 Nov. 1984

Nuclear Sciences and

3. G. Bidoglioj Characterization of Am(III) complexes with bicarbonate and carbonate ions at groundwater concentration level: Radiochem. Radioanal. Letters 53/1/45-60/1982

4. J. Hadermann, H. Schweingruber; Influence of speciation on the geospheric migration of radionuclides IAEA-SH-257/1P. IAEA-Vienna (1982)

TABLE I

Major Ion concentration in the solution used (Molar Units)

BRINE SOLUTION CLAY AQUIFER

Na+ 5.48 10-3

2.4 • 10-3 K+ 1.9 2.0 10-4 Mg2+ 1.1 . 10-3 1.3 . 10-4

Ca2+ 4.1 10-3 7.6 -5 . • 10_6 Fe(tot) 1.8 · 10_4 C12_ 5.48

10-3 1.8 · 10_6

SO 5'~5' 10-4

5.2 10_3 l:C02

10 • 3.4 • 10 pH 4 6 8.35

168

TABLE II

Stability Constant Values in the Speciation Modelling

Species ~

Am OH2+ 6.1

Am (OH); 9.48

Am C12+ 0.487

Am Cl+ 2 0.033

Am CO+ 3

5.81

Am (C03); 11.8

2-Am OH (C0

3)2, 15.9

Np020H 5.1

Np02Cl 0.4

Np02COi 4.6

169

r[- ,'l---7~f-:1~)'t~rf~'"-'_=--- --l[~]! () 50 100 150 :)00

,;ulumn h,nuth (111111)

Figure 1: -scanning profiles of the soil column loaded with 241-Am glass leachate at different bed volumes of brine solutions.

10

c c.

05

101..---

05

o

67 BV

I Po ? : I I I I I I 0 I I I I I ,

Y" 3U UV (/

I ,/ .. u ,r~ .. -.d ~~..-i-Q:-t-f>:i'~)"'o., ... =·~:-:: __ I n _____ ---r~---·

o 50 150 250 jlulcol.llillil lime (duy:;)

Figure 2 : breakthrough curve of the brine Am leachate ~he peaks shown Eorrespond to activity redistribution inside the column du­ring the scanning operation)

170

:L.O

05

RF t» Ae:ti nide.s in thE. S ~stem Na CL brine/sand catumn

10

Z11~p OXIC

DAYS

24JAm

119

f02.

Figure 3 breakthrough curves of Am and Np for the various situations studied

--

"-

-12 3

en 1 .-c en .2

o -9

log lAm I (M)

-11 -10 -9

Am (UI)

-~---k-_~ __ " ....

- 8 -7 -6

10gfNpJ (M)

Figure 4 : Adsorption isotherms of Am and Np (square points correspond to column influent concentrations)

171

% of Am chloro c::;r;,~lex:s

Figure 5 : Influence of pH and ~C02 on the amount of Am-chlorocomplexes 2+ +

AmCl + AmC12

in a brine solution (Cl-= 5.4M)

100

ItTI 5.4 M

Icoz I!f4 M

[/J XO t.Ll Np0 2011 U I t.Ll Np02C1 0... [/J

~

> i:l: 60 Z U.J ....l co ::J ....l 0 [/J 40 ~ I U.J NpO'2 ....l 0 ~

20

5

pH

Figure 6 : partitioning of Np(V) soluble species in a saturated brine solution

172

c

100 t \ t

80 oj

.. j .,/

",.,

I experim.nt.1 Vllu.

-,p

-- -.p' ~100 .. .p

-"---,~

20

I I I I I I I ,

I

Figure 7

D1 ... 0.1

. lilt" pcwlllily, ~m

retention of leached Americium versus filter porosity

10'6 r------------------------------------, .9 • 28 days -ct! '--g 10'7 () C o ()

~ 10'8

+

+ 14++ ~ -14' +

•• ; + ",. . r' •• '. +

++

• 80 days + 170 days

• •

•

10.11 L..--l\.---"_--l_--l._--L_-'-_-'-_-':-_-'-_-:! o 8 12 16 20

x. em (Column length)

Figure 8 : Americium column profile for three percolation experiments of different length

173

c.1 q, 'v ~ (.,f

Q,I ..... ...0

::s -... 0 \It

~ ~

~ 10-6 ,...---------------, C

_Q

e C~ 10-7 ~ 11 t = 170 days

§ \ u ••

E -10-8 .~ •

i ~... ---;:: lO-e ~e. t) ~ ••

e. -10-'0

10 -11 ':---'---',----'----.:_..L--:'-:-_-'---:-'=-_'--::'_ o 4 8 12 16 20

x. em (Column length)

Figure 9 : Americium filtration pattern, comparison between experimental and theoreti­cal data

1~~--~----~--------~------------~--__ ~

80

(,0

40

20

5 6 1- 8 pH

Figure 10: Distribution of Am soluble species as a function of pH for a chemical environment typical of a clay aquifer

174

LABORATORY EXPERIMENTS DESIGNED TO PROVIDE LIMITS ON THE RADIONUCLIDE SOURCE TERM FOR THE NNWSI PROJECT

V. M. Oversby and R. D. McCright

Lawrence Livermore National Laboratory Livermore, California 94550 USA

ABSTRACT

The Nevada Nuclear Waste Storage Investigations Project is investigating the suitability of the tuffaceous rocks at Yucca Mountain Nevada for potential use as a high level nuclear waste repository. The horizon under investigation lies above the water table, and therefore offers a setting that differs substantially from other potential repository sites. The unsaturated zone environment allows a simple, but effective, waste package design. The source term for r~dionuc1ide release from the waste package will be based on laboratory experiments that determine the corrosion rates and mechanisms for the metal container and the dissolution rate of the waste form under expected long term conditions. This paper describes the present status of laboratory results and outlines the approach to be used in combining the data to develop a realistic source term for release of radionuc1ides from the waste package.

, , RESUME

liThe Nevada Nuclear Waste Storage Investigations Project II est en train d'etudier 1a pertinence des rochers en tuf dans 1a montagne Yucca en Nevada pour son utilisation possible pour 1e stockage definitif des dechets radioactives. Le terrain, sous investigation, se situe au-dessus de la nappe aquifere et offre donc un emplacement qui differe notab1ement d'autres lieux de stockages possibles. Cette zone d'environnement non-sature nous permettrait de designer un conteneur simple et ~ la fois effectif. La I'source term" pour 1 'emission de radionuc1eides du conteneur des dechets se sera basee sur des experiences de laboratoire qui determineront 1e taux de corrosion et 1es mecanismes n~cessaires au conteneur meta11ique, ainsi que le taux de dissolution des dechets dans des conditions previsibles a long termes. Ce oapier decrit 1e statut present des resu1tats de laboratoire,et decrit 1 'approche qu'on uti1isera en combinant toutes les donnees pour developper une "source term" rea1iste aux fins de minimiser 1a relache de radionucleides du conteneur des dechets.

175

INTRODUCTION

The Nevada Nuclear Waste Storage Investigations Project is studying the tuffaceous rock formations at Yucca Mountain, Nevada, to determine the suitability of these rocks for construction of a high level nuclear waste repository. The horizon that is under investigation is the Topopah Spring Member of the Paintbrush Tuff, a welded, devitrified ash flow tuff. At Yucca Mountain, this unit lies in the unsaturated zone. The water table is approximately 150 m below the potential repository horizon in the center of the repository exploration block.

The unsaturated setting results in several conditions that are different from the more familiar circumstances for a repository designed for siting below the water table. Some of the important design considerations for the unsaturated zone include:

(1) The waste package will not be in contact with standing water under anticipated conditions;

(2) The waste package will not be subjected to lithostatic or hydrostatic load;

(3) The maximum temperature of liquid water that can contact the waste package is determined by the unconfined boiling point of water at the appropriate elevation (95°C at Yucca Mt.);

(4) Water influx will be limited to that available through downward infiltration (rainfall minus evapotranspiration);

(5) The waste packages will be in contact with air; (6) Container and waste form degradation will occur in air-steam and

dripping water environments.

WASTE PACKAGE ENVIRONMENT and WASTE PACKAGE DESIGN

The welded, devitrified sections of the Topopah Spring tuff consist of a fine grained intergrowth of quartz, cristoba1ite, and alkali feldspar, with a small proportion of phenocrysts. Phenocryst minerals are quartz, alkali feldspar, plagioclase, and biotite [lJ. Samples of water have not been obtained yet from the unsaturated zone tuff at the repository level; these samples will be obtained during the construction of the exploratory shaft. Pending availability of unsaturated zone water samples, the water from Well J-13, which is located several miles to the east of Yucca Mt., has been adopted as the reference ground water for NNWSI_experimenta1 work. The well is at a lower elevation than Yucca Mt. and the Topopah Spring tuff, which is below the water table at this elevation, is the major producing horizon for the well.

Experiments have been conducted to determine the potential changes in water chemistry due to interaction of the rock with J-13 water at elevated temperatures [2J. The expected upper limit temperature for liquid water in macroscopic openings at Yucca Mt. is 95°C; however, the capillary forces in pores may allow rock and water to equilibrate at somewhat higher temperatures. To account for this possibility, the rock-water experiments have been conducted at temperatures of 90, 120, and 150°C. A summary of the predicted steady-state water chemistry as a function of temperature is given in Table I. The bicarbonate content of J-13 water is approximately 140 ppm. Carbonate/bicarbonate following reaction with the rock depends on the partial pressure of carbon dioxide and the pH. The other anion components are those originally present in J-13 water: F = 2 mg/1, C1 = 7 mg/1, N03 = 9 mg/1,

176

and S04 = 18 mg/l. There is no significant source of these anions in the rock, so concentrations are independent of temperature.

TABLE I: Estimate of Steady State Water Chemistry for Topopah Spring Tuff and J-13 Water as a Function of Temperature

Concentration in mg/l at Element 25°C 90°C 120°C 150°C

Al 0.01 0.4 1.2 1 B 0.13 O. 1 O. 1 O. 1

Si 27 49 81 122 Ca 12.5 8 3.5 3 Mg 1.9 0.2 O. 1 O. 1

K 5. 1 9 9 9 Na 44 40 45 40

The maximum amount of water that can contact a waste package in a fixed time interval is a function of the infiltration rate and the area of repository rock that channels water flow past the waste package. For vertical emplacement of waste packages in boreholes in the floor of drifts, the maximum water drainage can be reasonably estimated by considering a unit cell of the drift to be associated with each waste package. For a drift width of 6.1 m and package borehole spacing along the drift floor of 5 to 8 m, the average drift area per package would be 40 sq.m. The water influx to that area would be 40 1 per year for an infiltration rate of 1 mm/y, the expected upper limit under normal conditions at Yucca Mt.[3J. By contrast, if only the water that flows directly downward onto the top of a borehole can enter that emplacement hole, the 1mm/y infiltration rate would deliver only 0.5 1 of water to the borehole. In order to investigate the effects of higher infiltration rates, we will use 1, 2, and 8 mm/y as cases for release rate scenario analysis.

Analysis of the waste package environment described above has led to the following reference waste package design for NNWSI. The design for spent fuel consists of a 70 cm. OD container fabricated from austenitic stainless steel [4J. The container thickness is 1 cm. An internal space frame will be used to separate intact fuel assemblies or consolidated bundles of fuel pins. The reference emplacement conditions are vertical emplacement in boreholes drilled in the floor of emplacement drifts. No- backfill or packing material is used around the waste packages. Reprocessed borosilicate glass waste forms from West Valley or from the Defense Waste Processing Facility would be placed in an overpack that would be similar to the spent fuel outer container. These reprocessed waste forms contain lower amounts of heat-producing radionuclides than spent fuel and are expected to be less complicated than spent fuel from the point of view of developing a waste package source term. The remainder of this paper will consider only the source term for spent fuel.

The reference waste package design for spent fuel assumes that the fuel is 10 years out of reactor and had an average burnup for PWR fuel of 33,000 MWd/MTHM. Figure 1 shows the thermal history for a waste package containing consolidated fuel pins from six PWR assemblies with a total thermal output of-3.3 kW. The emplacement conditions used in the calculation were vertical emplacement with 5 m package pitch and 47 m drift pitch. This produces an average thermal loading on emplacement of 48.4 kW per acre. The

177

peak temperature for the spent fuel would be approximately 320°C, while the canister surface experiences a peak temperature of somewhat less than 250°C. The host rock temperature for a distance of more than 1 meter around the borehole remains above 100°C for more than 1000 years under these emplacement conditions. Because the temperature surrounding the borehole is above the boiling point of water, the expected conditions for the waste package during the first 1000 years would be an atmosphere of air saturated with water vapor at the appropriate temperature. After the temperature drops below the local boiling point of water, liquid water might contact the container. For waste packages that contain lower thermal loading or are at the edge of the repository the time of first ingress of liquid water into the borehole would be expected to be earlier than for the conditions used in the example above.

3~rrrrrrrrrr~nnrnnnnnnrnnrn~nnHhnnnnMn~

-P 250 Canister O. D. -e

t i I- 200 -Host rock at Borehole

150·

100 ~~~~~~~~~~~~~~~~~~~~~~ o 100 200 300 400 500 600 700 800 900 1000

Time after emplacement (years)

Figure 1: Thermal history for reference waste package.

CONTAINER CORROSION RATES AND MECHANISMS

The reference container material for NNWSI waste packages is austenitic stainless steel. Several grades of stainless steel are under investigation, including 304L, 316L, 3l7L, 321, 347, and high-nickel stainless alloy 825. Of these materials, the laboratory work has emphasized 304L and

178

3l6L. Widespread industrial use of these grades indicates that they perform well in oxidizing conditions and, therefore, they are promising candidate materials for nuclear waste containers in an unsaturated tuff repository. The results of experiments conducted to determine the general corrosion rate of these materials immersed in J-13 water in the 50 to 100°C range indicate that the rate is essentially independent of temperature and the corrosion is uniformly distributed over the exposed surface area [5J. There was no significant difference in corrosion rates measured for 304L, 316L, and 317L in 50, 70, 80, 90 and 100°C J-13 water at exposure times of 3548 and 5000 hours. The grand average of data gave a corrosion rate of 0.19 +/- 0.04 ~m/y. If more recently obtained data for 7500 hours of exposure are included in the average, the rate is 0.15 +/- 0.036 ~m/y. The total range of measured corrosion rates in this series of experiments was 0.025 to 0.36 ~m/y.

Published [6J and unpublished results of testing of these materials in 100°C saturated steam (7500 exposure hours) gave a range of corrosion rates of 0.037 to 0.31 ~m/y and an average of 0.16 +/- 0.03 ~m/y. Oxidation rates ranging from 0.001 to 0.008 ~m/y have been measured in 150°C unsaturated steam (3800 exposure hours). As expected, the rates under "wet" steam conditions overlap those from the fully immersed conditions, while the rates obtained under "dry" steam conditions are substantially lower. Tests conducted in the presence of gamma radiation give similar results to those without radiation.

As discussed above, the exact temperature history of a waste package depends on its position in the repository layout and on the thermal loading of the package. In fact, these parameters are themselves interdependent. An exact description of the layout and package thermal loadings is not possible until the operations stage of repository development when shipping information from waste producers is available. To allow for the uncertainties in these parameters, we have chosen to use the 100°C air/steam data to represent the corrosion rate for higher temperatures. Since these data are comparable to the data for immersed conditions, we will use the grand average of the immersed corrosion data to estimate the waste package lifetime under the assumption that uniform corrosion is the service limiting degradation mode.

The only indications of localized corrosion that have been observed to date for 304L, 316L, and 317L occurred at 100°C and consisted of enhanced corrosion at the site of artificially induced crevices that were created underneath Teflon washers. The highest rate of preferential attack was about 0.3 ~m/yr after 7500 hours of exposure. No preferential attack was observed at 90°C and lower temperatures for these materials.

The other corrosion mechanism of concern for waste packages is intergranular stress corrosion cracking (IGSCC). Occurrence of IGSCC requires attainment of three concomitant conditions: (1) development of a continuously sensitized (Cr depleted) microstructure along grain boundaries; (2) the presence of a sufficiently aggressive corrosion environment to attack the Cr-depleted regions and initiate a crack; and (3) a stress level above a critical threshold \0 propagate the crack. This level is usually on the order of 70% of the yield strength. While waste package designs and container fabrication process specifications will ensure that the operating and residual stress levels are kept well below this threshold, it ;s instructive to determine whether the container alloy could become sensitized under the long term thermal conditions in the repository.

179

Figure 2 shows the temperature history for a waste package containing consolidated rods from BWR assemblies plotted as time vs liT (thermal data from [7J). The line for 304L sensitization was obtained by extrapolation of data obtained at higher temperatures (450-7500e). This particular material had a carbon content (0.028%) near the top of the permitted range and was heavily cold-worked [8J. Both of these conditions enhance susceptibility to sensitization. The offset of the waste package cooling curve from the sensitization line suggests that long-term, low-temperature sensitization of even heavily cold-worked material should not be of concern. Additional conservatism can be incorporated into the container material selection, however, as illustrated by the line for non cold-worked 3l6LN stainless steel [9J. Additions of nitrogen and molybdenum in the 3l6LN retard the diffusion of carbon in austenitic structures and therefore reduce the susceptibility to sensitization. Figure 2 further illustrates that experiments can be performed at 250 to 4000e for one to a few years in duration to confirm the predictions on sensitization made from the higher temperature extrapolations discussed above.

The data available on corrosion in the environment anticipated at Yucca Mt. indicate that uniform corrosion is the most likely mechanism for container degradation. Extrapolation of the data for the average uniform corrosion rate (0.15 um/yr) to obtain the time necessary to corrode through 1 cm of metal gives a predicted package lifetime of 67,000 years. erevi~e corrosion under saturated steam conditions at 1000e would reduce this lifetime somewhat. The duration of experiments is presently limited to under one year (7500 h = 312.5 days) and more data is required to determine that increased corrosion rates do not occur following prolonged exposure. Present results, however, indicate that failure of the metal container in times less than 10,000 years will be an unlikely event. Potential causes of premature breach of the metal container could include defective closure welds that are not detected during weld inspection and portions of the container being highly stressed due to some off-normal fabrication condition. These are considered to be low probability conditions; however, the consequences of these conditions are being evaluated as part of the metal corrosion testing program.

SPENT FUEL DEGRADATION RATE

Light water reactor fuel generally consists of U02 pellets contained in Zircaloy cladding. A small percentage of fuel is clad in stainless steel. For simplicity, we will confine discussion to Zircaloy clad fuel. A similar treatment will apply to stainless steel clad fuel.

At the end of its useful life, most spent fuel cladding is intact; a small percentage of cladding will have developed a breach during reactor service. Fuel designs and reactor operating conditions used during the 1960 l s led to development of pin hole or crack defects in several percent of the pins. More recent designs and operating conditions result in far fewer defects in the cladding. The defect population for fuel presently in use has been estimated to be about 0.1% or less [lOJ.

We are investigating the corrosion rates and mechanisms for Zircaloy under expected disposal conditions for a repository in tuff. Preliminary analysis indicates that the dominant corrosion mechanism will be stress corrosion cracking. Degradation by uniform corrosion or pitting corrosion is expected to be minor [llJ. Experiments are currently under way to measure the

180

rate of uniform and non-uniform corrosion of irradiated Zircaloy. Results of these experiments will be used to determine the rate of defect generation in the spent fuel cladding population. Experiments are also being conducted on spent fuel oxidation rates since fuel oxidation through small cladding defects might result in an increased rate of cladding degradation.

"More conservative metallurgical case"

316 In SS, low carbon high nitrogen,

solution annealed

1

-en ... C1l Q)

\ > -..... c \ Q)

E Q)

10 ~ (.)

~ Q. \ E Q)

\ ... Q) ..... .....

\ C1l Q)

E \ I-

100 \ \ \ \

Sensitized microstructure ..

Cooling curve for reference SF waste canister - 10-yr old waste, 3.42 kW power load - No packing material - Vertical emplacement

"Worst metallurgical case" 304L ss with rather high carbon

heavi Iy -cold-worked

Non-sensitized microstructure ~

1000L-4-0LO~-L-U~3~O-0~--25~O~L--2~0-O--~--1~5-0--~-----1~0-O~

Temperature (OC)

Figure 2: Relationship between thermal history of emplaced nuclear waste container and long-term sensitization in austenitic stainless steels.

181

Leaching of radionuc1ides from spent fuel can occur only after the cladding is breached and water or air can contact the fuel. In most cases, contact of fuel with water will occur through small defects in the cladding. In some cases, the cladding may split and expose the fuel pellets directly to water. To cover the possible geometric configurations for fuel-water contact, we are conducting spent fuel dissolution studies that use four fuel conditions: (1) bare fuel plus the cladding from which the fuel was removed, (2) a fuel segment approximately five inches long, end capped with water tight fittings, and no defects induced in the cladding, (3) specimen similar to number 2, but with a laser drilled hole in the cladding, and (4) specimen similar to number 2, but with a slit machined in the cladding. Specimen 2 provides an indication of the surface contamination of the cladding. Such contamination may occur during reactor service, pool storage at the reactor, or during preparation of the specimens for testing. Details of the experimental procedure are given in reference [12J.

The first series of spent fuel tests were conducted in deionized water at ambient hot cell temperatures and used Turkey Point PWR fuel. Solution samples (generally 10 m1) were taken periodically and fresh deionized water added to the test vessel to return the volume to 250 m1. The test was run for approximately 9 months, temporarily terminated to allow transfer of the fuel test specimens to new reaction vessels, and then restarted in fresh deionized water. Results for uranium in solution samples for the first sixty days of the first cycle of the test are shown in Figure 3, along with results for a similar series of tests using J-13 water and H.B. Robinson PWR spent fuel. The solution concentration that would be produced if one part in 105 of the inventory present in the test were dissolved is shown for reference for each test. Details of results for the deionized water tests are contained in reference [13J and for the J-13 water tests in reference [14J.

The bare fuel sample in deionized water showed an initial pulse of U release followed by a rapid decrease of U in the solution. Subsequent testing showed that the U was not in true solution, but was contained in particles larger than 0.4 urn in size. The concentration of other elements in the solutions, such as Pu, indicated that the initial pulse of release was due to suspension of very fine particles of fuel that rapidly settled out and plated out onto the reaction vessel surface. Analysis of acid rinse solutions of the test vessels after initial termination of the test gave compositions that indicated plate-out of material with an inventQry of elements that would be expected in bulk spent fuel. When the tests were restarted with fresh deionized water, another pulse of release was observed, but it was substantially smaller than the first pulse. Release from specimens with inauced cladding defects did not show an initial pulse of release, and resulted in solution concentrations that were one or two orders of magnitude lower than for the bare fuel. Eventually the particulate U settled out and final solution concentrations were similar to those found for the intact control sample. Settling time for the initial test was over 8 months, while in the restarted test settling occurred after only 4 months. The amount of uranium present in true solution or in particles smaller than 0.4 urn was only a few ppb in all tests in deionized water.

Results for testing of bare fuel in J-13 water also show an initial pulse of release that is similar in magnitude to that observed in deionized water. In this case, the uranium becomes comp1exed with carbonate in the solutions and is present as a true solution species. The uranium

182

i ..... r

,02 .----.,----.-,.-,-----.,-----.,.---.-----,..-----

rf"

.Q------------~ E------O--.-- .

- DEIONIZED WATER TESTS --- J·13 WelL WATER TESTS o BARE FUEL o SLIT DEFECT I:::. HOLES DEFECT • UN DEFECTED

10~L------L----~~----~----~~----~----~----~ o ~ DAYS

H(Olll4Ol '!D.2

Figure 3: Uranium in unfiltered solutions from testing of PWR spent fuel. See text for details.

183

concentration decreases slowly as the test progresses. This may be due to removal of some of the initial "pulse" material as analytical samples are taken that is not replaced by further spent fuel reaction. The control sample (marked undefected) shows U at levels between that of the laser hole and slit defect samples. This indicates a small amount of residual contamination on the control sample, amounting to about 5 micrograms of U. The J-13 water tests are still in progress and will follow a test plan similar to that for the deionized water tests. A second series of tests using Turkey Point fuel in J-13 water is also in progress. Tests using H. B. Robinson fuel and J-13 water under higher temperature conditions are planned.

Some radioactive species in spent fuel are concentrated in the gap between the pellet and cladding and/or on the grain boundaries of the fuel. These species became segregated during reactor operation and-lie in regions that allow them to be released to solution at a rate faster than the matrix fuel dissolves. Cesium, iodine, and possibly technetium, show an initial release rate that is high compared to the U release and that correlates with the content of fission gas in the pellet-cladding gap [13,15J. The amount of cesium and similar species that is contained in the highly mobile fraction is usually less than 1% of the inventory in the pin [lOJ; the remainder is dispersed through the fuel matrix and should be released at the same rate as the matrix dissolves. For the tests discussed above, Cs release from bare fuel was essentially instantaneous and corresponded to about 0.2% in the Turkey Point tests in either solution and about 0.6% in the H. B. Robinson tests. In the deionized water tests the release of Cs from slit and hole defects was slower but eventually reached the same levels as the bare fuel initial release. On restarting the deionized water tests, an additional rapid release was seen; however, the amount of release was smaller than for the first contact with water.

A third source of radioactivity in spent fuel exists in the metal components of the fuel assembly. Neutron activation products are formed in the fuel cladding and in the metal assembly spacers and grids. For steel components, the corrosion rates will probably be similar to those determined as part of the container corrosion testing. This will be checked by a limited series of experiments on fuel assembly parts. The main long-lived radionuclides contained in the steel components will be nickel isotopes.

The most significant radionuclide pre~ent in the cladding is C-14. It is formed by the reaction N-14 (n,p) C-14. Experiments conducted at temperatures of about 275°C have shown that approximately 0.2% of the inventory of C-14 from a fuel assembly can be released into an air atmosphere within about one month [16J. The process involved in the release is thought to be removal of carbon from the oxidized skin of the Zircaloy cladding by reaction of the oxygen in the atmosphere with the carbon in the cladding oxidation layer to release C02. The proposed mechanism will be checked by experiments conducted under controlled conditions using defueled cladding. Further release of C-14 from the Zircaloy is much slower than the initial release from the oxidized layer, and will probably be controlled by the rate of oxidation of the Zircaloy. Since this rate is very low at low temperatures, release of C-14 will probably be slow after the initial oxide layer carbon has been removed.

The experiments described above result in a source term for spent fuel that will be composed of at least four components that are released at

184

different

A separate components

rates. The parts will consist of (1) Elements whose release is controlled by matrix dissolution; (2) Elements present in part in the pellet-cladding gap; (3) Elements contained in stainless steel spacers and grids; (4) Elements contained in the fuel cladding. dissolution/degradation rate will be determined for each of these of the source term.

WASTE PACKAGE SOURCE TERM DEVELOPMENT

Data from the experiments described above will be used to develop the model for release rate of radionuclides from the waste package as a function of time. A distribution function will be developed for container breach that includes both expected conditions and off-normal conditions. Data presently available suggest that breach of a container earlier than 50,000 years after disposal would not occur under anticipated conditions.

The container breach-time distribution will give the starting point for radionuclide release from the waste package. The first radionuclide released would be C-14 from the oxidized outer layer of the Zircaloy cladding. This release does not require contact of water with the fuel. Following container breach, water might enter the container. The geometric description of the breached container and the expected water flow rate will be used to determine the appropriate water to fuel ratio to be used in modelling the breached container. Present data indicate that infiltration rates of 0.1 to 8 mm/y should cover the range of likely conditions.

A waste package with one or more cracks is the most likely form of a breached container. Depending on the location of the cracks, infiltrating water might collect to some depth in the container. For the purposes of initial source term modelling, we will assume that two cracks exist with both cracks near the top of the container. The volume of a container ;s less than 1800 1, while the weight of spent fuel in the container is about 3100 kg. More than half of the volume will be occupied by the fuel, so the maximum water to fuel ratio would be about 0.3 l/kg. The spent fuel dissolution studies described above use water to fuel ratios of 3 to 10 l/kg. It is unlikely that we will be able to conduct experiments at a water/fuel ratio as low as 0.3; however, we will establish dissolution rates over a range of water/fuel ratios so that extrapolation to the appropriate value can be made.

The source term will include an estimate of the rate of degradation of the Zircaloy cladding. Spreading the breach of cladding over a range of time has a large effect on the release rate of elements contained in the pellet-cladding gap. If the cladding fails at a uniform rate over 2000 years, a rapid release of 1% of the iodine on breach of cladding would become a popu­lation release rate of 0.5 parts in 105 of the iodine inventory due to that mechanism. The total release rate of iodine would then be found by summing the parts due to the rapid release fraction and that due to matrix dissolution.

The Zircaloy, stainless steel, and spent fuel dissolution and degradation rates will then be used to determine the concentrations of radionuclides in the water contained within a breached waste package. This information, when combined with the water infiltration rate, will then provide the volume of water containing radionuclides that is displaced from the breached container annually, and the radioactive inventory contained in that

185

water. This calculation will provide the source term for radioactive release from the waste package.

REFERENCES

[lJ Bish, D.L. et a1.: "Preliminary Stratigraphic and Petrologic Characterization of Core Samples from USW-G1, Yucca Mountain, Nevada", Los Alamos Nat1. Lab. LA-8840-MS, 1981.

[2J Oversby, V. M.: "Reaction of the Topopah Spring Tuff with J-13 Water at 120°C", Lawrence Livermore Nat1. Lab. UCRL-53574, 1984.

[3J Montazer, P. et a1.: " Drilling, Instrumentation, and Preliminary Testing Results of Borehole USW UZ-1 Penetrating Thick Unsaturated Tuffs, Nevada Test Site", U.S. Geological Survey, USGS-WRI-84-x, 1984(in prep.).

[4J 0'Nea1, W.C. et a1.: "Prec10sure Analysis of Conceptual Waste Package Designs for a Nuclear Repository in Tuff ", Lawrence Livermore Nat1. Lab. UCRL-53595, 1984.

[5J Glass, R.S. et a1.: "Electrochemical Determination of the Corrosion Behavior of Candidate Alloys Proposed for Containment of High Level Waste in Tuff", Lawrence Livermore Nat1. Lab. UCID-20174, 1984.

[6J McCright, R.D. et al.: "Selection of Candidate Canister Materials for High-Level Nuclear Waste Containment in a Tuff Repository", Lawrence Livermore Natl. Lab. UCRL-89988, 1983.

[7J Hockman, J. N. and W. C. 0'Nea1: "Thermal Modeling of Nuclear Waste Package Designs for Disposal in Tuff", Lawrence Livermore Nat1. Lab. UCRL-89820 Rev. 1, 1984.

[8J Fox, M.J. and R. D. McCright: "An Overview of Low Temperature Sensitization", Lawrence Livermore Nat1. Lab. UCRL-15619, 1983.

[9J Mulford, R.A., et a1.: "Sensitization of Austenitic Stainless Steels II: Commercial Purity Alloys", Corrosion 39, 132, 1983.

[lOJ Woodley, R.E.: liThe Characteristics oY-Spent LWR Fuel Relevant to its Storage in Geologic Repositories", Westinghouse Hanford Co. HEDL-TME 83-28, 1983.

[11J Rothman, A.J.: "Potentia1 Corrosion and Degradation Mechanisms of Zirca10y Cladding on Spent Nuclear Fuel in a Tuff Repository", Lawrence Livermore Nat1. Lab. UCID-20172, 1984.

[12J Wilson, C.N.: "Test plan for Spent Fuel Cladding Containment Credit Tests", Westinghouse Hanford Co. HEDL-TC-2353-2, 1983.

[13J Wilson, C.N.: "Resu1ts from NNWSI Series 1 Spent Fuel Leach Tests", Westinghouse Hanford Co. HEDL TME 84-30, ~984.

[14J Wilson, C.N. and V.M. Oversby: "Radionuc1ide Release from PWR Fuels in J-13 Water", Lawrence Livermore Nat1. Lab. UCRL-91464, 1984 (in prep.).

[15J Johnson, L.H. et a1.: "Re1ationship Between Fuel Element Power and the Leaching of 137Cs and 129 1 from Irradiated U02 Fue1" Proc. of the ANS Topical Meeting on Fission Product Behavior and Source Term Research, Snowbird, Utah, July 1984.

[16J Van Konynenburg, R.A. et a1: "Behavior of Carbon-14 in Waste Packages for Spent Fuel in a Repository in Tuff", Lawrence Livermore Nat1. Lab. UCRL-90855, 1984.

*Work performed under the auspices of the U.S. Department of Energy by the Lawrence Livermore National Laboratory under contract number W-7405-ENG-48.

186

THE POTENTIAL EFFECTS OF RADIATION ON THE SOURCE TERM IN A SALT REPOSITORY

G. L. McVay, L. R. Pederson, W. J. Gray, and D. Rai Pacific Northwest Laboratory

ABSTRACT

The escape of radionuclides from a salt repository depends upon release from the engineered barrier system (source term) and upon an accident scenario which allows exit from the salt media. Several factors can affect the source term. These factors include the interactive effects of waste package materials, the particular accident scenario, and radiolysis effects. The current discussion will focus upon radiolysis effects.

Gamma radiolysis can have its main effect by enhancing metallic hard barrier corrosion and by generating sodium metal colloids within the surrounding crystalline salt. When the irradiated salt is then exposed to brine, local basic pH excursions can occur. This can have significant impact upon the source term by causing changes in both leach rates and solubilities.

At longer times, after breaching the hard barriers, alpha radiolysis of the brine can occur. Alpha radiolysis can cause high gaseous pressures, very oxidizing conditions in the brine, and low pH values. Both the highly oxi­dizing brine (high Eh) and low pH can result in as much as three orders of magnitude increase in actinide solubilities .

. Recent research is described on the effects of gamma radiolys;s on crystal­line salt core samples and the subsequent effects of reacting the irradiated salt with brine. Alpha radiolysis effects on saturated brines and upon actinide solubility are also discussed.

(Abstract only available)

187-188

THE CHEMISTRY OF NUCLEAR FUEL (UOZ

) DISSOLUTION UNDER WASTE DISPOSAL VAULT CONDITIONS

D.W. Shoesmith and M.G. Bailey Research Chemistry Branch

Atomic Energy of Canada Limited Whiteshell Nuclear Research Establishment

Pinawa, Manitoba, Canada ROE 1LO

ABSTRACT

The basic chemistry of UOZ dissolution is considered under condi­tions typical of a nuclear fuel waste disposal vault, with particular emphasis on redox conditions and the effect of groundwater constituents, oxidizing agents and water radiolysis. The anodic processes that occur during the oxidative dissolution of UOZ have been established on the basis of electro­chemical and X-ray photoelectron spectroscopic experiments. Using similar techniques, the kinetic effects of potential groundwater species, such as bicarbonate/carbonate, phosphate, and chloride, have been studied.

The effect of potential oxidizing agents is not so well character­ized. A number of interrelated reactions with UO

Z' involving dissolved

oxygen, hydrogen peroxide, and the products of water radiolysis, have been identified and discussed. The directions in which our studies are being extended are discussed.

RESUME

La dissolution du UOZ a ete examineedu point de vue chimique, en payant une attention particuliere aux conditions oxido-reductrices, a l'effet des constituants des eaux souterraines, aux agents oxidants et a la radiolyse de l'eau. Les processus anodiques survenant lors de la dissolution oxidative du U02 ont ete etablis en se basant sur des experiences electrochimiques et sur la spectroscopie photoelectronique au rayons-x. En utilisant des tech­niques similaires, l'effet cinetique des especes presentes dans les eaux souterraines, tel que les ions carbonate/bicarbonate, phosphate et chlorure, a ete etudie.

L'effet d'agents oxidants n'est pas si bien characterize. de reactions etroitement imbriquees, impliquant l'oxygene dissout, le d'hydrogene et les produits de la radiolyse de l'eau a ete identifie. direction future de ces etudes est discutee.

189

Un nombre peroxyde

La

1. INTRODUCTION

The development of a model to predict the rate of release of radio­nuclides from irradiated nuclear fuel under waste disposal vault conditions is a task requiring a detailed knowledge of the chemistry of the host matrix (UO ), and how it is affected by irradiation and the various groundwater con~itions to which it may be exposed. To achieve such an understanding, an integrated experimental approach involving thermodynamic, kinetic and leaching studies is being followed. Some of the details of this approach have been reviewed recently [1].

The physical and chemical properties of irradiated fuel depend on its irradiation history. Physical changes that occur during irradiation include grain growth, cracking, and the production of open and closed porosity due to the formation of fission gas bubbles at grain boundaries. These factors combine to produce an inhomogeneous distribution of radionuclides throughout the fuel. Radionuclides have been identified in a number of specific sites, e.g., in void spaces, such as the fuel/sheath gap, cracks in the fuel pellets, and inter-pellet gaps; at UOZ grain boundaries; and within UOZ grains.

The rate of release of the various radionuclides from these loca­tions has been the subject of many leaching studies [Z-8]0 However, since over 90% of the radionuclide inventory is located within the UOZ grains, the chemistry of dissolution of the UO matrix is of major importance in the development of any radionuclide release model. Some of the factors that will exert a major influence on the chemistry of UO include redox conditions [9-13], groundwater composition [11,13], radiation effects, temperature, physical effects such as fuel pellet erosion and break-up [4], and fuel/ container interactions [14].

In this paper we review some of our basic studies on UOZ dissolution, with particular emphasis on redox conditions and the effect of groundwater constituents, oxidizing agents and water radiolysis. Since the control of redox conditions is essential to any study of the dissolution of UOZ' we have used a variety of electrochemical methods and X-ray photoelectron spectro­scopy (XPS) to characterize the surface films formed on UOZ under a variety of redox conditions.

2. EXPERIMENTAL

All of our experiments have been performed on a UOZ electrode fabricated from an unirradiated fuel pellet, as previously described [4]0 The details of our electrochemical and XPS procedures and analyses have been published previously [4,15-17]. The cell used in our a-radiolysis experi­ments has also been described previously [18]0

30 RESULTS AND DISCUSSION

3.1 Redox Conditions

On the basis of an extensive series of electrochemical and XPS experiments [4,15-17,19], we have established the following reaction scheme for the oxidative dissolution of UOZ!

190

(a) ~) (d) (e) (f) D0

2 + D02+x

+ D02•33 + D02• S

+ D02• 67 + D0

3

(c) l. (g) (1) D02+ + D03·xH2O 2

The film growth process (a) is limited to a few atomic layers [16,20], the film being thicker the higher the pH. No major crystallographic rearrangement is required to convert D02+x to D02.33 «b) in scheme (1». Film growth probably occurs via 02- insertion at interstitial sites in the fluorite lattice, and proceeds to thicknesses of about 4 to 8 nm [16,20].

The dissolution process (c) (to yield DO;+, almost certainly com­plexed or hydrolyzed) appears to start once a surface composition of D02.33 is achieved, irrespective of solution composition [20], or whether oxidation is promoted electrochemically [16], or by dissolved oxidants [21]. In step (d), the onset of dissolution is accompanied by a slow surface recrystallization process, from D02.33 with a fluorite lattice, through D02.S' to D02.67 with an orthorhombic lattice. This second, slower stage of oxidation, and the disso­lution process, are envisaged as proceeding via a common intermediate (uranyl­type species) produced by the oxidative dissolution of D0

2•33

•

The final step, (f), to D03 , is only observed in alkaline solutions [19]. The phase D03.xHZO is formed by the precipitation of dissolved DO~+ from locally supersaturated solutions, and is generally only observed at very anodic potentials (~+300 mV vs SCE) [16], or after extensive dissolution at lower potentials [4].

From the waste management point of view, these results show that a degree of oxidation can be tolerated prior to the onset of dissolution of the D02 matrix. We have attempted to determine the minimum electrode potential for the onset of oxidative dissolution. A value in the region 0 to +SO mV (vs SCE) was obtained from coulometric experiments [17]. This could be taken as a minimum redox condition, below which D02 should be a stable host matrix for radionuclides.

3.2 Groundwater Composition

Although the relative importance of the dissolution/film formation reactions changes slightly, our electrochemical results show that the mechan­ism of oxidative dissolution of D02 remains essentially unchanged in the pH range S < pH < 10, the range expected in groundwaters. These results are in agreement with those of Nicol and Needes [22], who showed that the pH had no effect on the steady-state current vs potential curves for the anodic disso­lution of D02 in this pH range.

However, the effect of potential groundwater anions, such as carbon­ate and phosphate, is more complicated. The early stages of D02 oxidation, when no dissolution is occurring (scheme (1), steps (a) and (b», are unaffected by the presence of complexing anions [20,23]. Our coulometric experiments, aimed at determining the redox potential at which dissolution first occurs, suggest that carbonate does not affect this potential. At more anodic potentials, our electrochemical experiments show a substantial effect for the complexing anion, in qualitative agreement with the earlier results of

191

Nicol and Needes [24]. Our inability to observe an effect due to carbonate at low anodic potentials may be due, in part, to the presence of dissolved atmos­pheric C02 in our supposedly carbonate-free solution. Consequently, a redox potential of 0 to +50 mV (vs SCE) appears to represent a genuine, minimum, redox condition for the onset of oxidative dissolution, even in the presence of strong uranyl ion complexing agents such as carbonate.

At more anodic potentials, dissolution proceeds through a surface intermediate, U02C03 in the case of carbonate, and through an unidentified uranyl phosphate (possibly U02HP04) in the case of phosphate. For carbonate solutions, an acceleration of the dissolution reaction is observed for [C03]total ~ 10-4 mol.dm-3 , and the chemical dissolution of the U02C03 layer tends to partially control the dissolution kinetics for [C03]total > 10-3

mol·dm-3• For phosphate, only a small effect on the dissolution rate is observed unless [P04]total > 5x10-2 mol.dm-3• The uranyl phosphate film is much less soluble than the corresponding carbonate film and completely controls the dissolution kinetics under sufficiently oxidizing conditions (> 325 mV vs SCE).

On the basis of these and similar experiments for other anions, we can define the following conditions under which we would expect various anions to affect the dissolution rate of U0 2 :

(1)

(2)

(3)

3.3

Carbonate/bicarbonate will accelerate the oxidative dissolution of U0 2 for [C03]total ~ 10-4 mol·dm-3• Since concentrations in the range

10-3 to 10-2 mol·dm- 3 are expected in groundwaters in the Canadian Shield [13], carbonate is likely to affect the chemistry of U02 in a waste vault.

Since phosphate concentrations less than 10-5 mol.dm- 3 are expected in groundwater [13] and concentrations > 5x10-2 mol o dm-3 are required to affect the dissolution reaction, phosphate is not expected to have a significant effect on the oxidative dissolution of U02•

-1 -3 No effect of chloride ion up to 10 mol·dm has been observed, suggest-ing that the saline groundwaters sometimes encountered in the Canadian Shield [13] pose no extra threat to U02 fuel stability, unless radiolysis of chloride solutions produces oxidizing ~pecies.

Effect of Oxidizing Agents

The electrochemical studies described above are limited to the anodic reactions occurring on U02, the cathodic reaction being relegated to a subsidiary electrode elsewhere in the experimental system. To determine the kinetics of the corrosion of U02, we also need to know the mechanism of the cathodic reaction. A number of cathodic reactions are possible, since oxidizing agents may be introduced to the waste vault either by transport in groundwater, or by radiolytic decomposition of water within the vault.

We have chosen to study U02 corrosion in solutions containing dis­solved oxygen as an example of an oxidizing reagent that may be transported into the waste vault, though it is possible to envisage many scavenging mech­anisms for the removal of dissolved oxygen, such as preferential reactions with components of the buffer material and with the metal containers [14].

192

Oxidizing conditions at the fuel surface could also be produced by the radiolytic decomposition of water, viz.,

Consequently, we have been studying the effect of H2

0Z

and the actual a-radiolysis of water on UO Z corrosion [18].

(a) Dissolved Oxygen

(Z)

Figure 1 shows the U02 corrosion potential recorded in oxygen­saturated 0.1 mol'dm- 3 NaCI04 (pH = 9.5). The data points are from 13 separate experiments, showing good reproducibility. The potential rises rapidly (~ 1 min) to ~ -400 mV from -ZOOO mV, the potential at which the electrode was cathodically cleaned. Over the next 6 to 8 hours, the corrosion potential rises to ~ +30 mV and a thin film is observed (by cathodic stripping voltammetry) to grow on the UOZ- A limiting thickness of 6±1 nm is achieved. For longer times, the corrosion potential slowly drifts to ~ +90 mV, but no further film thickening is observed.

It appears that, as in the electrochemical case, an initial surface oxidation occurs, followed by dissolution from a U02+x surface film. It is tempting to assume that oxidation to the U02.33 stage (step (b), scheme (1)) has occurred and that dissolution is proceeding by the oxidative dissolution of this layer. However, this hypothesis requires confirmation by XPS and dissolution rate measurements.

The reduction of oxygen on the U02 surface (the accompanying cathodic reaction), via the overall reaction

(3)

appears to be first order in oxygen, indicating no oxygen dissociation on the U02 surface prior to electron transfer. On the basis of the electrochemically determined Tafel slopes and the pH dependence of the reduction currents, Needes and Nicol [25] have proposed that the first electron transfer reaction is rate-determining. This reaction is then followed by a sequence of three further electron transfer reactions to produce OH-.

A major feature of this mechanism is that the reduction does not appear to proceed via a peroxide intermediate, unlike 02 reduction at many other metal surfaces [26]. This suggests that there will be fundamentally different mechanisms for the reactions of U02 with 02 and with H202 (produced by water radiolysis). We are presently studying the corrosion of U02 over a wider range of pH, temperature and solution composition.

(b) Hydrogen Peroxide

Figure 2 shows the corrosion potential of U02 as a function of time after the addition of H2020 The two concentrations emEtoyed are close to those of dissolved 02 in air-saturated ([02] = 2.66x10 mol o dm- 3) and oxgyen-saturated ([02] = 1.27x10-3 mol-dm- 3) solutions. A surface oxidation of U02 occurs, similar to that observed in dissolved oxygen. However, the rate of oxidation is much faster in H202 than in 02, which may be due to a

193

reaction involving OH radicals, as discussed below. Thus, the corrosion potential achieves a value of +85 mV (vs SCE) in rv 5 min in a solution con­taining 10-3 mol·dm- 3 of H202 compared with rv 24 h for a similar concentration of dissolved 02. .

Figure 3 shows the U02 corrosion potential as a function of H202 concentration, as well as the potential achieved at a Pt surface in the same solutions. For the U02 electrode, the potential is effectively independent of the H202 concentration for concentrations ~ 10-2 mol·dm- 3 , whereas for the Pt electrode, the potential is constant over the concentration range 5x10-3 < [H202] < 2x10- 1 mol·dm-3 • It is well known that Pt catalyzes the decomposi­tion of H202 [27] according to the overall reaction

-+ (4)

The near invariance of the potential with H202 concentrations < 10-2 mol·dm-3

suggests the decomposition reaction (4) is occurring at both the Pt and U02 electrodes. Examination of the two half-cell reactions involved in the peroxide decomposition reaction,

H202 + 2H+ + 2e ---'>0. 2H2O (5) ..--

H202 ~

°2 + 2H+ + 2e (6) ..--

shows that, provided the reaction mechanism does not change, the shift in potential with H202 concentration for the two half-cell reactions will be equal but of opposite sign. Consequently, the mixed potential would not be expected to change, as observed in Figure 3. That H202 decomposition does occur on U02 in this potential region is confirmed by the rotating ring-disk experiments of Needes and Nicol [25].

-2 -3 For [H202] > 10 mol·dm ,the shift of the U02 electrode potential to more anodic potentials suggests that further oxidation of the U02 is occurring. Our XPS results [21] show that the U02 surface is predominantly U02.33 at low concentrations of H202' the phases U0 2 •5 and U02.67 being formed at the more anodic potentials achieved in the higher concentration region. These observations agree with our electrochemical results described above, and are consistent with the rotating ring-disk results of Needes and Nicol [25], which show that the formation of surface phases on the U02 leads to a switch from predominantly peroxide decomposition, reaction (4), to the oxidative dissolution of U0 2 , as the electrode potential is decreased.

A comparison of these results with those obtained in dissolved 02 shows that two distinctly different reaction pathways are possible, depending on the peroxide concentration. At low concentrations, H202 decomposition will lead to 02 formation, and the subsequent rate of U02 dissolution by reaction with oxygen would be expected to be slow. At higher concentrations, direct reaction of H202 with U0 2 will occur and dissolution is expected to be accel­erated. Our results suggest that there is a threshold concentration of H202 (rv 5x10- 2 mol.dm-3) above which U02 dissolution is enhanced. Hence, for H202 produced by water radiolysis, we might expect a dose-rate threshold for enhanced U02 dissolution (and radionuclide release), since the H202 concentra­tion is determined by the dose rate. So far our work has been confined to the pH range 9 to 10.

194

(c) Alpha radiolysis

Figure 4 shows the corrosion potential of a D02 electrode as a func­tion of time in argon-degassed solutions with the electrode in close proximity (~ 30 ~m) to a-particle sources of various strengths. In the presence of weak a sources, the potential increases quickly from -2000 to -1000 mV and then slowly drifts to about -200 mV. This behaviour is similar to that obtained in the absence of an ex source and can be considered to represent "nominally reducing" conditions, the small amount of D02 oxidation being attributable to residual traces of dissolved oxygen. In the presence of the 686 ~Ci* ex source, the change in potential from "nominally reducing" (up to 2 h) to oxidizing (> 10 h) reflects the buildup of a radiolysis products as a function of time. The change in potential is slow compared with the changes observed in H202 (Figure 4) and oxygen-saturated (Figure 3) solutions. The potential of +70 to +75 mV (vs SCE) is similar to that achieved in H202 and 0z­containing solutions, demonstrating that oxidizing conditions are achieved at the D02 surface as a consequence of water radiolysis. The surface composition has yet to be established by XPS.

Dp to ~ 30 h (at which time the experiment was terminated), the D02 potential remains in the region of ~ +70 mV (vs SCE). This suggests that the H202 concentration in the small volume of solution between the electrode and the a source does not build up to concentrations ~ 10-2 mol.dm-3 , where we would expect an anodic shift in potential (see Figure 3) and an extensive attack on the D020 At a potential of +70 mV (vs SCE), we would expect H202 decomposition via reaction (4). It is also possible that radical reactions occur at the D02 surface, although no evidence for such reactions presently exists. Experiments in which the volume of solution between the electrode and the a source, and hence the concentration of radiolysis products, are varied [18] support these conclusions.

If H202 decomposition occurs on D02' then the main mechanism of D02 oxidation and dissolution will be via reaction with 02. As shown above, this reaction is significantly slower than the reaction of D02 with H202, and would not be expected to lead to as severe an attack. The relative importance of the decomposition of H202 to D02 dissolution remains to be determined.

So far our experiments have been confined to durations of :is }O h. For this time period, the two weaker ex sources -(4.7 ~Ci and 100 ~Ci), which produce ex fluxes close to those expected to occur at the surface of irradiated fuel after 500-a storage, showed no tendency to oxidize the D02 surface. It is tempting to interpret this switch in behaviour from "nominally reducing" to oxidizing, as the strength of the ex source increases, as a dose-rate effect. However, it is most likely due to the cell design [18], which allows the transport of radiolysis products out of the solution volume between the D02 electrode and the ex source. At the low dose rates obtained with the two weaker a sources, these transport losses will be more significant in deter­mining the concentration of radiolysis products at the D02 surface.

We are presently developing a model to calculate the expected yields of radiolysis products, using published models for the ex radiolysis of water,

*1 Ci 376 Bq

195

and taking into account the transport losses of radiolysis products. Also, in experiments now in progress, we are attempting to use the gold-plated surface of the ex source as an electrode, in the hope of analyzing the yield of radiol­ysis products by electrochemical techniques. By comparing the measured values with those calculated using the model, we hope to be able to measure the effect of dose rate.

Finally, all the data presented in this section on oxidizing agents are for pH 9 to 10. Our preliminary experiments at other pH values suggest a much more aggressive attack on U02 at lower pH values, especially in H202 solutions. Obviously, this effect of pH, plus the effect of radiation on solution species such as chloride and carbonate, must be studied in detail before a more complete picture of U02 dissolution under oxidizing conditions will be possible.

4. CONCLUSIONS

The anodic processes occurring on U02 under electrochemical condi­tions, and in the presence of oxidants such as dissolved 02 and H202 have been elucidated.

The corrosion potential of U02 has been measured in solutions containing dissolved 02, H202 , and in the presence of an ex source. In all cases oxidizing conditions are achieved at the U02 surface, but the rate of oxidation is much faster in H202 solutions than in solutions containing dissolved 02' In the presence of H202 , or the ex radiolysis products of water, H202 decomposition and U02 dissolution occur simultaneously at the U02 surface, the relative importance of each reaction being determined by the H202 concentration.

5. REFERENCES

1. Wikjord, A.G., Shoesmith, D.W., Goodwin, B.W. and Sargent, F.P.: "Canadian Research on High-Level Nuclear Waste Products and Processes", Proc. Int. Conf. Radioactive Waste Management, p. 383, Winnipeg, 1982, Canadian Nuclear Society (1982).

2. Johnson, L.H., Burns, K.I., Joling, H.H. and Moore, C.J.: "The Dissolu­tion of Irradiated U02 Fuel Under Hydrothermal Oxidizing Conditions", Atomic Energy of Canada Limited Technical Record*, TR-128 (1981).

3. Johnson, L.H.: "The Dissolution of Irradiated U0 2 Fuel in Groundwater", Atomic Energy of Canada Limited Report, AECL-6837 (1982).

4. Johnson, L.H., Shoesmith, D.W., Lunansky, G.E., Bailey, M.G. and Tremaine, P.R. : "Mechanisms of Leaching and Dissolution of U02 Fuel", Nuc!. Tech. 2§., 238 (1982).

5. Johnson, L.H.: "The Effect of Fuel Power Rating on the Release of 137Cs and 1291 During Leaching", Proc. 10th Information Meeting of the Nuclear Fuel Waste Management Program, p. 275, Atomic Energy of Canada Limited Technical Record*, TR-166 (1981).

196

6.

7.

8.

Johnson! L.H., Burn~, K.I., Joling, H.H. and Moore, C.J.: "Leaching of 137Cs , 34Cs and 12 I from Irradiated U0 2 Fuel", Nucl. Tech. 21, 470 (1983) •

Johnson, L.H., Stroes-Gascoyne, S., Shoesmith, D.W., Bailey, M.G. and Sellinger, D.M.: "Leaching and Radiolysis Studies on U02 Fuel", proc. 3rd Spent Fuel Workshop, KBS Technical Report, 83-76, L. Werme (ed.), Stockholm, 1984.

Johnson, L.H., Stroes-Gascoyne, So, Chen, J.D., Attas, M.Eo, Sellinger, D.M. and Delaney, H.G,,: "The Relationship Between Fuel Element Power and the Leaching of b7Cs and 1291 from Irradiated Fuel".

9. Tremaine, P.R., Chen, J.D., Wallace, G.J. and Boivin, W.A.: "Solubility of Uranium(IV) Oxide in Alkaline Aqueous Solutions to 300oC", J. SolIn Chern. lQ.(3), 221 (1981).

10. Ryan, J .L. and Rai, D.: "The Solubility of Uranium(IV) Hydrous Oxide in Sodium Hydroxide Solutions Under Reducing Conditions", Polyhedron l(9), 947 (1983).

11. Langmuir, D.: "Uranium Solution-Mineral Equilibria at Low Temperatures with Applications to Sedimentary Ore Deposits", Geochim. Cosmochim. Acta !tl, 547 (1978). .

12. Lemire, R.J. and Tremaine, P.Ro: "Uranium and Plutonium Equilibria in Aqueous Solutions to 200oC", JoChem. Eng. Data Q, 361 (1980).

13. Paquette, J. and Lemire, R.J.: "A Description of the Chemistry of Aque­ous Solutions of Uranium and Plutonium to 2000 C Using Potential-pH Diagrams", Nucl. Sci. and Eng. J1., 26 (1981).

14. Shoesmith, D.W. and Bailey, M.G.: "The Effects of Container-Corrosion and Alpha-radiolysis of Water on the Corrosion of U02", to appear in Proc. AECL Workshop on Container Corrosion, Ottawa, 1983.

15. McIntyre, N.S., Sunder, S., Shoesmith, D.W. and Stanchell, F.W.: "Chemi­cal Information from XPS-Applications to the Analysis of Electrode Surfaces", J. Vac. Sci. Technol. ]&(3), 714 (1981).

16. Sunder, S., Shoesmith, D.W., Bailey, M.G., Stanchell, F.W. and McIntyre, N. S.: "Anodic Oxidation of U02' Part 1. Electrochemical and X-Ray Photoelectron Spectroscopic Studies in Neutral Solutions", J. Electro­anal. Chern. 130, 163 (1981).

17. Sunder, S., Shoesmith, D.W., Bailey, M.G. and Wallace, G.J.: "Mechanism of Oxidative Dissolution of U02 Under Waste Disposal Vault Conditions", Proc. Int. Conf. Radioactive Waste Management, Winnipeg, 1982, Canadian Nuclear Society, Toronto, 1982.

18. Bailey, M.G., Johnson, L.H. and Shoesmith, D.W.: "Effects of Alpha­radiolysis of Water on the Corrosion of U02", accepted for publication in Corros. Scio

197

19. Sunder, S., Shoesmith, D.W., Bailey, M.G. and Wallace, G.J.: "Anodic Oxidation of U02. Part II. Electrochemical and X-Ray Photoelectron Spectroscopic Studies in Alkaline Solutions", J. Electroanal. Chem. 150, 217 (1983).

20. Shoesmith, D.W., Sunder, S., Bailey, M.G., Wallace, G.J. and Stanchell, F.W.: "Anodic Oxidation of U02. Part IV. X-Ray Photoelectron Spectro­scopic and Electrochemical Studies of Film Growth in Carbonate-Containing Solutions", Applic. of Surf. Sci., in press.

21. Sunder, S., Shoesmith, D.W., Bailey, M.G. and Wallace, G.J.: unpublished results.

22. Nicol, M.J. and Needes, C.R.S.: "The Anodic Dissolution of Uranium Dioxide - I in Perchlorate Solutions", Electrochimica Acta 20, 585 (1975) •

23. Shoesmith, D.W., Sunder, So, Bailey, M.G. and Owen, DoG.: "Anodic Oxidation of U02. Part III. Electrochemical Studies in. Carbonate Solutions", Proc. 5th Into Sympo on Passivity, p. 125, Bombannes, France, 1983, M. Froment (ed.), Elsevier, Amsterdam, 19830

240 Nicol, M.J. and Needes, CoR.S.: "The Anodic Dissolution of Uranium Dioxide - II in Carbonate Solutions", Electrochimica Acta~, 1381 (1977) •

25. Needes, C.RoS. and Nicol, Mojo: "A Study of Some Redox Reactions at a U02 Surface", Nat. Inst o Meto, Repub. S. Afr. Rep. No. 7073, 19730

260 Schiffrin, D.J.: "The Electrochemistry of Oxygen", Specialist Periodical Reports, The Royal Society of Chemistry, London, Vol. 8, 1983.

27. Prabhu, V.G., Zarapkar, L.R. and Daneshwar, R.G.: "Electrochemical Studies of Hydrogen Peroxide at a Platinum Disc Electrode", Electro­chimica Acta~, 725 (1981).

*Unrestricted, unpublished report, available from SDDO, Atomic Energy of Canada Limited Research Company, Chalk River, Ontario KOJ IJO.

198

~ +200,r---------~-----------r----------~----------~ w u •• _ •••••• (/) • ,wi .-... - .. ~ 0 • ..., 1' ... '-

::> 1"'-' E • "\,-,,

...... -200 , .. c-- ---...J I U- -~ -: .. I- .,

~ -400 w b Cl... -600 z o 2-800 a:: a:: 8

FILM GROWTH

I

I

TIME(h)

CONSTANT FILM ___ ~Hl~~~E~~ _____ •

DISSOLUTION

I

10

Figure 1. Corrosion potential of a U02 electrode in an oxygen-saturated solution ([02] = 1.27x10-3 mol.dm-3) containing 001 mol o dm-3

NaCl04 (pH = 9.5). Data points are from 13 individual experiments. Electrode potential = -2000 mV at t = O.

100 w .-e--e e-U

80 o-o-o-o-~;:::.e---(/) p-- ."e-VI > f ' :> 60 t

e

E ,

40 l ...J I e <[

i= 20 o I z

1 I w I-0 0 Cl... II z

-20 0 (/) o I 0

-40 11 a:: a:: 0

-60 u 2 4 6 8 10 12 14 16 18

TIME (min)

Figure 20 Corrosion potential of a U02 electrode in a N2-degassed 001 mol·dm-3 NaCl04 solution (pH = 9.5) containing hydrogen peroxide. (--~-) [H202] = 2x10-4 mol·dm- 3 ; (-~-) [H202] = 10-3 mol o dm-3• Electrode potential = -2000 mV at t = O.

199

W +250 u (f)

~ +200 :> E -1 +150 <r i= ~ +100 5 a. w o o 0:: ~ U W -1 W

+50

o

-50 Pt

""lIir---e--e-_e-e-

10-3 10-2

[H202JEn01. dm-3) Figure 3. Steady-state corrosion potential of a UOZ electrode (-~-), and

the potential at a Pt electrode (-~-), as a function of HZOZ concentration, in 0.1 mol·dm-3 NaC104 solution (pH = 9.5).

w u (f)

(/)

> ::>

o

E -200

Figure 4.

I

TIME(h)

10

Corrosion potential of a UOZ electrode in an Ar-degassed 0.1 mol.dm-3 NaC104 solution (pH = 9.5) (-0-0-) UOZ electrode ~ 30 ~m from a 4.7 ~Ci Z41Am a source; (-.-.-) ~ 30 ~m from a 100 ~Ci Z41Am a source; (0) ~ 30 ~m from a 686 ~Ci Z10po a source; (e) repeat experiment with 686 ~Ci a source.

200

WORKSHOP SUMMARY AND CONCLUSIONS

This workshop provided a unique opportunity to examine the role and underlying technical basis for the source terms used in calculating the long-term performance of nuclear waste repositories. It allowed a brief, but extremely revealing, exchange of information on modeling approaches, fundamental mechanisms, and experimental results by the participants. The representation of programs from many different countries allowed the devel­opment of an international perspective on the role of source terms. While in many cases there are different approaches and emphases in the various countries, numerous similarities and common understandings were revealed in the presentations and discussions.

The workshop first concentrated on the appropriate role of source terms in radionuclide isolation and on the need for definition and restrictions on the performance of the source term. In all cases, the knowledge of source-­term performance clearly provides (1) a measure of the integrity of the waste form (high-level waste or spent nuclear fuel) in the repository envi­ronment and (2) definition of the rates of radionuclide release into the groundwater system. Different requirements are apparent in the programs represented by workshop participants. In the U.S., the regulatory agencies have stated specific performance objectives for the lifetime of the engi­neered barriers that provide containment of radioactivity and for the rate of release, as a function of the inventory present at 1000 years, from the engineered systems into the groundwater. The requirements in other countries are based largely on minimizing radionuclide release into the biosphere, but engineered systems that virtually guarantee containment for tens of thousands of years are proposed in many countries.

The spatial location of the source term received considerable atten­tion. No common definition or specification was agreed upon. It was clear that different repository configuration and modeling approaches will require different definitions of the source term. However, it was believed that the definition is determined by the modeling method used and that it will in most cases be the interface where waste package modeling stops and transport calculations begin.

The description of the source term and the conditions that determine its characteristics are directly tied to the assumptions about the prevail­ing geohydrologic conditions in the repository. Conditions resulting from unlikely, disruptive events considered credible over extremely long times will also influence the source--term descriptions. These descriptions are generally presented in terms of scenarios for groundwater flow and waste package degradation. A wide range of scenarios was presented. Hard rock repositories (granite, basalt, tuff, etc.) are generally characterized as having small fluxes of water traveling either in the rock matrix or in frac­tures. Additional differences in defining scenarios are introduced by con­sidering saturated or unsaturated repository conditions. A repository in

201

salt provides different mechanisms for source-term determination. The potential for thermally induced migration of in situ brine is being con­sidered. The creep deformation of the rock has the potential not only to seal passageways and prevent waste-water contact but also to provide a mechanism for moving brine, initially present because of flooding of the underground cavities, to overlying aquifers. Finally, the inherent solubil­ity of salt provides the means for various degrees of contact of brine with the waste form if unsaturated brine sources are available near the reposi­tory.

In all cases, the mechanisms that primarily determine the characteris­tics of the source term can be placed into four categories: the first deals with the amount of groundwater available to interact with the waste package and the geochemical condition of that water after modification by the ther­mal and radiation conditions in the vicinity of the waste. In the second category, the controlling mechanisms are those that determine the time and extent of degradation of the engineered barriers protecting the waste form. A third category of mechanisms deals with the dissolution of the waste form, including the selective degradation of various constituents and the specia­tlon and quantity of individual radionuclides that are released. The last category deals with the host rock near the waste package, where the dominate mechanisms determining the source term involve the transport of radionuclides along the flow path and the competing processes for retarding or stimulating movement with respect to the groundwater flux.

While the source term modeling and experimental studies generally fit into these broad categories, a number of specific mechanisms were revealed that warrant additional evaluation. These include (1) the potential for alpha radiolysis of local groundwater that might create a "redox front," consisting of an oxidizing region that moves with the radionuclides in the groundwater system, (2) the quality and effects of gas produced from radioly­sis and corrosion, (3) corrosion and failure mechanisms for canister mate-­rials that allow for water entry into the waste packages and the subsequent exit of solutions containing radionuclides, (4) the potential for formation of colloids and the chemical and physical mechanisms for their retardation in the groundwater system, and (5) the synergistic effect of each of these phenomena acting together over long periods of time.

Finally, the experimental basis for parameters defining the source term was examined. The evolution of repository concepts for specific host rock appears sufficiently advanced to allow an emphasis on experimental simulations of geochemical and hydrologic environments that would be present in a specific medium. It was believed such experiments should be encouraged. Further, the modeling of repository performance has also developed to the point that specific parameter needs and ranges can be specified. Thus, it is appropriate that interaction between experimentalists and modelers be encouraged and increased beyond that currently evident in most programs.

The process of combining the conceptual framework used by the modelers, the environmental conditions and controlling mechanism being defined by analysts, and the existing data base and experimental capabilities of the experimentalist was initiated by this workshop. It is hoped that additional interaction will provide elaboration on specific individual effects and further integration of these fundamental approaches into a comprehensive basis for modeling repository performance.

202

NEA WORKSHOP ON THE SOURCE TERM FOR RADIONUCLIDE MIGRATION

Albuquerque, New Mexico November 13-15, 1984

PARTICIPANTS LIST

M. APTED Pacific Northwest Laboratory P.O. Box 999 Richland, Washington 99352 USA

Tel: (509) 375-2156 Tlx: 152874

A. AVOGADRO Commission of European Communities Joint Research Centre 21020 Ispra (VA) ITALY

Tel: 332/789954 Tlx: 380042 EUR I

J. BRAITHWAITE Organization 6312 Sandia National Laboratories P.O. Box 5800 Albuquerque, New Mexico 87185 USA

Tel: (505) 846..,.1186

J. CLEVELAND U.S. Geological Survey M.S. 412 P.O. Box 25046 Denver, Colorado 80225 USA

Tel: (303) 236-5043

F.T. EWART UKAEA Harwell, Oxford UNITED KINGDOM

Tel: 0235 24141 Tlx: 83135

F. GERA ISMES via T. Taramelli 14 00197 Rome ITALY

Tel: (06) 804351-5 Tlx: 614615 FOR ISMES

J. HIGGO Imperial College of Science

and Technology Department of Chemistry South Kensington, London SW7 2AZ UNITED KINGDOM

Tel: (01) 589 5111 Ext. 538 or 4384

Tlx: 261503

T. HUNTER Sandia National Laboratories P. o. Box 5800 Albuquerque, NM 87185 USA

Tel: (505) 844-9160

G. JOHANSSON National Institute of Radiation

Protection Box 60204 S-10401 stockholm SWEDEN

Tel: 08-24 40 Tlx: 11771 safared S

J. KERRISK Los Alamos National Laboratory Mail stop 6787 Los Alamos, New Mexico 87545 USA

Tel: (505) 667-3348

K. KRAUSKOPF Stanford University Geology Department B-73 Stanford, California 94305 USA

Tel: (415) 497-3325

203

NEA Workshop Albuquerque, New Mexico November 13-15, 1984

(cont. )

F. LANZA Commission of European Communities Joint Research Centre 21020 Ispra (VA) ITALY

Tel: 332/780131 Tlx: 380042 EUR I

I. McKINLEY EIR CM5303 WUrenlineen SWITZERLAND

Tel: 056 99 24 20 Tlx: 53714 eir ch

G. McVay Pacific Northwest Laboratory P.O. Box 999 Richland, Washington 99352 USA

Tel: (509)·375-3762 Tlx: 15-2874

A. MAYO University of Colorado-

Colorado Springs P.O. Box 7970 Colorado Springs, Colorado 80933 USA

Tel: (303) 593-3223

A. MULLER Organisation for Economic

Co-operation and Development 38, boulevard Suchet 75016 PARIS FRANCE

Tel: 524 82-00 Tlx: 630.668 AEN/NEA

1. NERETNI EKS Department of Chemical Engineering Royal Institute of Technology S-10044 Stockholm 70 SWEDEN

Tel: 08 7878229 Tlx: 53 178 ethbi ch

204

V. OVERSBY Lawrence Livermore National Laboratory L-206 Livermore, California 94550 USA

Tel: (415) 423-2228 Tlx: 910-386-8339

S. PASQUINI CEN Cadarache DRDD/l:!EGC B.P. i 13115 Saint Paul Les Durance FRANCE

Tel: (42) 25 39 45 Tlx: CEACA 440 678 F

E. PELTONEN Technical Research Centre of Finland Nuclear Engineering Laboratory PL 169 SF-00181 Helsinki 18 FINLAND

Tel: (90) 648 931 Tlx: 122972 vttha sf

G. RAINES Battelle Memorial Institute ONWI Battelle Project Management Division 505 King Avenue Columbus, Ohio 43201 USA

Tel: (614) 424-7832 Tlx: 24-5454

L. RICKERTSEN Roy F. Weston, Inc. 2301 Research Boulevard Rockville, Maryland 20850 USA

Tel: (301) 963-5214 Tlx: 83-5348

NEA Workshop Albuquerque, New Mexico November 13-15, 1984

(cont.)

P. SALTER Rockwell Hanford P.O. Box 800 Richland, Washington 99352 USA

Tel: (509) 373-4200 Tlx: 373-2908

D. SHOESMITH Atomic Energy of Canada Research

Company Chemistry Branch Whiteshell Nuclear Research

Establishment pinawa, Manitoba ROE 1LO CANADA

Tel: (204) 753-2311 Tlx: 07-57553

M. SNELLMAN Technical Research Centre of

Finland Reactor Laboratory Otakaari 3 A SF-02150 Espoo 15 FINLAND

Tel: 90-4561 Tlx: 122972 VTTIN SF

P. SOO Nuclear Waste Management Division Brookhaven National Laboratory Upton, New York 11973 USA

Tel: (516) 282-4094 Tlx: 282-2547

N. STELTE Gesellschaft fUr Strahlen und

Umweltforschung 3300 Braunschweig Theodor-Heuss-Strasse 4 FRG

Tel: 8012 286 Tlx: 952 865 iftta

205

E. WICK Division of Waste Management U.S. Nuclear Regulatory Commission Washington, D.C. 20555 USA

Tel: (301) 427-4111 Tlx: 908142 NRC - BHD WSH

A. WIKJORD Atomic Energy of Canada

Research Company Geochemistry and Applied

Chemistry Branch Whiteshell Nuclear Research

Establishment Pinawa, Manitoba ROE 1LO CANADA

Tel: (204) 753-2311 Tlx: 07--57553

T. YANAGIDA Japan Atomic Energy Research

Institute H.L.W. Management Lab Tokai, Naka, Ibaraki JAPAN

Tel: 0292-82-6180 Tlx: 3632340JTOKAIJ

DISTRIBUTION

M. APTED Pacific No~thwest Labo~ato~y P.O. Box 999 Richland, Washington 99352 USA

A. AVOGADRO Commission of Eu~opean Communities Joint Resea~ch Cent~e 21020 Isp~a (VA) ITALY

J. W. BENNETT (RW-22) Office of Geologic Reposito~ies U.S. Depa~tment of Ene~gy Fo~~estal Building Washington, DC 20585

N. E. CARTER Battelle Columbus Labo~ato~y Office of Nuclea~ Waste Isolation 505 King Avenue Columbus, OH 43201

J. CLEVELAND U.S. Geological Su~vey M.S. 412 P.O. Box 25046 Denve~, Colo~ado 80225 USA

COMMISSION OF THE EUROPEAN COMMUNITIES

200 Rue de la Loi B-1049 B~ssels BELGIUM

DOCUMENT CONTROL CENTER Division of Waste Management U.S. Nuclea~ Regulato~y Commission Washington, DC 20555

F.T. EWART UKAEA Ha~ell, Oxfo~d

UNITED KINGDOM

J. J. FIORE (RW-22) P~og~am Management Division Office of Geologic Reposito~ies U.S. Depa~tment of Ene~gy Fo~restal Building Washington, DC 20585

M. W. FREI (RW-23) Enginee~ing & Licensing Division Office of Geologic Reposito~ies U.S. Department of Ene~gy Fo~~estal Building Washington, DC 20585

R. W. GALE (RW-40) Office of Policy, Integ~ation, and Out~each

U.s. Depa~tment of Ene~gy Fo~~estal Building Washington, DC 20585

F. GERA ISMES via T. Ta~amelli 14 00197 Rome ITALY

W. M. HEWITT P~og~am Manage~

Roy F. Weston, Inc. 2301 Resea~ch Blvd., 3~d F1oo~ Rockville, MD 20850

J. HIGGO Impe~ial College of Science

and Technology Depa~tment of Chemist~y South Kensington, London SW7 2AZ UNITED KINGDOM

G. JOHANSSON National Institute of Radiation P~otection

Box 60204 S-10401 Stockholm SWEDEN

J. KERRISK Los Alamos National Labo~ato~y Mail stop 6787 Los Alamos, New Mexico 87545 USA

K. KRAUSKOPF Stanfo~d Unive~sity

Geology Depa~tment B-73 Stanfo~d, Califo~nia 94305 USA

206

F. LANZA Commission of European Communities Joint Research Centre 21020 Ispra (VA) ITALY

S. A. MANN Manager Crystalline Rock Project Office u.s. Department of Energy 9800 South Cass Avenue Argonne, IL 60439

A. MAYO University of Colorado-

Colorado Springs P.O. Box 7970 Colorado Springs, Colorado 80933 USA

I. McKINLEY EIR CM5303 Wilrenlineen SWITZERLAND

G. McVAY Pacific Northwest Laboratory P.O. Box 999 Richland, Washington 99352 USA

A. MULLER (10) Organisation for Economic

Co-operation and Development 38, boulevard Suchet 75016 PARIS FRANCE

J. O. NEFF Manager Salt Repository Project Office U.S. Department of Energy 505 King Avenue Columbus, OH 43201

I. NERETNIEKS Department of Chemical Engineering Royal Institute of Technology S-10044 Stockholm 70 SWEDEN

O. L. OLSON Manager Basalt Waste Isolation Project Office U.S. Department of Energy Richland Operations Office Post Office Box 550 Richland, WA 99352

V. OVERS BY Lawrence Livermore National

Laboratory L-206 Livermore, California 94550 USA

S. PASQUINI CEN Cadarache DRDD/BECC B.P. 1 13115 Saint Paul Les Durance FRANCE

E. PELTONEN Technical Research Centre of

Finland Nuclear Engineering Laboratory PL 169 SF-001S1 Helsinki 18 FINLAND

G. RAINES Battelle Memorial Institute ONWI Battelle Project Management

Division 505 King Avenue Columbus, Ohio 43201 USA

L. RICKERTSEN Roy F. Weston, Inc. 2301 Research Boulevard Rockville, Maryland 20850 USA

B. C. RUSCHE (RW-1) Director Office of Civilian Radioactive

Waste Management U.S. Department of Energy Forrestal Building Washington, DC 20585

207

SAIC-T&MSS LIBRARY Science Applications

Inte~national, Corpo~ation

2950 South Highland D~ive Las Vegas, NY 89109

P. SALTER Rockwell Hanfo~d P.O. Box 800 Richland, Washington 99352 USA

D. SHOESMITH Atomic Ene~gy of Canada Resea~ch

Company Chemist~y B~anch Whiteshell Nuclea~ Resea~ch

Establishment Pinawa, Manitoba ROE lLO CANADA

M. SNELLMAN Technical Resea~ch Centre of Finland Reacto~ Labo~ato~y

Otakaa~i 3 A SF-02150 Espoo 15 FINLAND

P. SOO Nuclea~ Waste Management Division B~ookhaven National Labo~ato~y Upton, New Yo~k 11973 USA

RALPH STEIN CRW-23) Office of Geologic Reposito~ies U.S. Depa~tment of Ene~gy Fo~~estal Building Washington, DC 20585

N. STELTE Gesellschaft fu~ St~ahlen und Umweltfo~schung

3300 B~aunschweig Theodo~-Heuss-St~asse 4 FRG

TECHNICAL INFORMATION CENTER Roy F. Weston, Inc. 2301 Resea~ch Bouleva~d, Thi~d Floo~

Rockville, MD 20850

D. L. VIETH Di~ecto~

Waste Management P~oject Office U.S. Depa~tment of Ene~gy Post Office Box 14100 Las Vegas, NY 89114

E. WICK Division of Waste Management U.S. Nuclear Regulato~y Commission Washington, D.C. 20555 USA

A. WIKJORD Atomic Ene~gy of Canada

Resea~ch Company Geochemist~y and Applied

Chemist~y B~anch

Whiteshell Nuclea~ Resea~ch Establishment

Pinawa, Manitoba ROE lLO CANADA

T. YANAGIDA Japan Atomic Ene~gy Resea~ch

Institute H.L.W. Management Lab Tokai, Naka, Ibaraki JAPAN

208

Internal Distribution:

6300 R. W. Lynch 6310 T. O. Hunter (10) 6310 NNWSICIi' 6311 L. W. Scully 6311 L. Perrine (2) 6312 10' • W. Bingham 6312 J. W. Braithwaite 6312 M. S. Tierney 6313 T. E. Blejwas 6314 J. R. Tillerson 6315 S. Sinnock 6332 WMT Library (20) 6430 N. R. Ortiz 3141 C. M. Ostrander (5) 3151 W. L. Garner (3) 8024 M. A. Pound DOE/TIC (28)

(3154-3, C. H. Dalin)

209-210